@@ -1582,14 +1582,14 @@ Data cleaning
method scale(strategy="standard")
-
+
Applies one of sklearn's scalers. Non-numerical columns are ignored (instead
of raising an exception). See the Scaler class.
method clean(prohibited_types=None, strip_categorical=True, maximum_cardinality=True,
- minimum_cardinality=True, missing_target=True, encode_target=None)
-
+ minimum_cardinality=True, drop_duplicates=False, missing_target=True, encode_target=None)
+
Applies standard data cleaning steps on the dataset. Use the parameters
to choose which transformations to perform. The available steps are:
@@ -1605,7 +1605,7 @@ Data cleaning
method impute(strat_num="drop", strat_cat="drop", min_frac_rows=None, min_frac_cols=None, missing=None)
-
+
Impute or remove missing values according to the selected strategy. Also removes
rows and columns with too many missing values. The imputer is fitted only on the
training set to avoid data leakage. Use the missing attribute to customize what
@@ -1615,7 +1615,7 @@
Data cleaning
method encode(strategy="LeaveOneOut", max_onehot=10, frac_to_other=None)
-
+
Perform encoding of categorical features. The encoding type depends on the
number of unique values in the column:
@@ -1632,7 +1632,7 @@ Data cleaning
method prune(strategy="z-score", method="drop", max_sigma=3, include_target=False, **kwargs)
-
+
Prune outliers from the training set. The definition of outlier depends
on the selected strategy and can greatly differ from one each other.
Ignores categorical columns. Only outliers from the training set are pruned
@@ -1642,7 +1642,7 @@
Data cleaning
method balance(strategy="ADASYN", **kwargs)
-
+
Balance the number of samples per target class in the target column. Only the training
set is balanced in order to maintain the original distribution of target classes
in the test set. See Balancer for a description of
@@ -1667,7 +1667,7 @@
Feature engineering
method feature_generation(strategy="DFS", n_features=None, generations=20, population=500, operators=None)
-
+
Use Deep feature Synthesis or a genetic algorithm to create new combinations
of existing features to capture the non-linear relations between the original
features. See FeatureGenerator for
@@ -1677,7 +1677,7 @@
Feature engineering
method feature_selection(strategy=None, solver=None, n_features=None,
max_frac_repeated=1., max_correlation=1., **kwargs)
-
+
Remove features according to the selected strategy. Ties between features with
equal scores are broken in an unspecified way. Also removes features with
too low variance and finds pairs of collinear features based on the Pearson
@@ -1730,21 +1730,21 @@
Training
method run(models, metric=None, greater_is_better=True, needs_proba=False, needs_threshold=False,
n_calls=10, n_initial_points=5, est_params=None, bo_params=None, bagging=0)
-
+
Runs a DirectClassifier instance.
method successive_halving(models, metric=None, greater_is_better=True, needs_proba=False,
needs_threshold=False, skip_runs=0, n_calls=0, n_initial_points=5,
est_params=None, bo_params=None, bagging=0)
-
+
Runs a SuccessiveHalvingClassifier instance.
method train_sizing(models, metric=None, greater_is_better=True, needs_proba=False,
needs_threshold=False, train_sizes=np.linspace(0.2, 1.0, 5), n_calls=0,
n_initial_points=5, est_params=None, bo_params=None, bagging=0)
-
+
Runs a TrainSizingClassifier instance.
Example
diff --git a/docs/API/ATOM/atomloader/index.html b/docs/API/ATOM/atomloader/index.html
index 53a094fd4..1d930d414 100644
--- a/docs/API/ATOM/atomloader/index.html
+++ b/docs/API/ATOM/atomloader/index.html
@@ -572,7 +572,7 @@ ATOMLoader
function ATOMLoader(filename, data=None, transform_data=True, verbose=None)
-
+
Load a class instance from a pickle file. If the file is a trainer that
was saved using save_data=False, you can load new data into it. For
atom pickles, you can also apply all data transformations in the
diff --git a/docs/API/ATOM/atommodel/index.html b/docs/API/ATOM/atommodel/index.html
index 4f33d2b37..9701aa59e 100644
--- a/docs/API/ATOM/atommodel/index.html
+++ b/docs/API/ATOM/atommodel/index.html
@@ -581,8 +581,8 @@
ATOMModel
function ATOMModel(estimator, acronym=None, fullname=None, needs_scaling=False)
-
-Convert an estimator to a model that can be ingested by ATOM.
+
+Convert an estimator to a model that can be ingested by atom.
| Parameters: |
diff --git a/docs/API/ATOM/atomregressor/index.html b/docs/API/ATOM/atomregressor/index.html
index 28f877c9a..e6be6c8ce 100644
--- a/docs/API/ATOM/atomregressor/index.html
+++ b/docs/API/ATOM/atomregressor/index.html
@@ -606,7 +606,7 @@ ATOMRegressor
class atom.api.ATOMRegressor(*arrays, y=-1, n_rows=1, test_size=0.2, logger=None,
n_jobs=1, warnings=True, verbose=0, random_state=None)
-
+
ATOMRegressor is ATOM's wrapper for regression tasks. Use this class to easily apply
all data transformations and model management provided by the package on a given
dataset. Note that contrary to sklearn's API, an ATOMRegressor instance already
@@ -1039,7 +1039,7 @@
Utility methods
method add(transformer, columns=None, train_only=False)
-
+
Add a transformer to the current branch. If the transformer is
not fitted, it is fitted on the complete training set. Afterwards,
the data set is transformed and the transformer is added to atom's
@@ -1082,7 +1082,7 @@
Utility methods
method apply(func, column)
-
+
Transform one column in the dataset using a function (can
be a lambda). If the provided column is present in the dataset,
that same column is transformed. If it's not a column in the
@@ -1111,7 +1111,7 @@
Utility methods
method automl(**kwargs)
-
+
Uses the TPOT package to perform
an automated search of transformers and a final estimator that maximizes
a metric on the dataset. The resulting transformations and estimator are
@@ -1130,7 +1130,7 @@
Utility methods
method canvas(nrows=1, ncols=2, title=None, figsize=None, filename=None, display=True)
-
+
This @contextmanager allows you to draw many plots in one figure. The default
option is to add two plots side by side. See the user guide
for an example use case.
@@ -1168,7 +1168,7 @@ Utility methods
method delete(models=None)
-
+
Removes a model from the pipeline. If all models in the pipeline are removed,
the metric is reset. Use this method to remove unwanted models or to free
some memory before saving the instance.
@@ -1185,7 +1185,7 @@ Utility methods
method distribution(column=0)
-
+
Compute the KS-statistic
for various distributions against a column in the dataset. Missing values are ignored.
@@ -1215,7 +1215,7 @@
Utility methods
method drop(columns)
-
+
Drop columns from the dataset.
Note
@@ -1236,7 +1236,7 @@
Utility methods
method export_pipeline(model=None)
-
+
Export atom's pipeline to a sklearn's Pipeline. Optionally, you can add a model
as final estimator. If the model needs feature scaling and there is no scaler in
the pipeline, a StandardScaler
@@ -1263,7 +1263,7 @@
Utility methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -1282,7 +1282,7 @@ Utility methods
method report(dataset="dataset", n_rows=None, filename=None)
-
+
Create an extensive profile analysis report of the data. The report is rendered
in HTML5 and CSS3. Note that this method can be slow for n_rows > 10k.
@@ -1319,13 +1319,13 @@ Utility methods
method reset_predictions()
-
+
Clear the prediction attributes from all models.
Use this method to free some memory before saving the trainer.
method save(filename=None, save_data=True)
-
+
Save the instance to a pickle file. Remember that the class contains the complete
dataset as attribute, so the file can become large for big datasets! To avoid this,
use save_data=False.
@@ -1348,7 +1348,7 @@ Utility methods
method save_data(filename=None, dataset="dataset")
-
+
Save the data in the current branch to a csv file.
@@ -1367,7 +1367,7 @@ Utility methods
method scoring(metric=None, dataset="test", **kwargs)
-
+
Print all the models' scoring for a specific metric.
@@ -1386,7 +1386,7 @@ Utility methods
method stacking(models=None, estimator=None, stack_method="auto", passthrough=False)
-
+
Add a Stacking instance to the models in the pipeline.
@@ -1423,19 +1423,19 @@ Utility methods
method stats()
-
+
Print basic information about the dataset.
method status()
-
+
Get an overview of the branches, models and errors in the current instance.
This method prints the same information as atom's __repr__ but will also
save it to the logger.
method voting(models=None, weights=None)
-
+
Add a Voting instance to the models in the pipeline.
@@ -1495,14 +1495,14 @@ Data cleaning
method scale(strategy="standard")
-
+
Applies one of sklearn's scalers. Non-numerical columns are ignored (instead
of raising an exception). See the Scaler class.
method clean(prohibited_types=None, strip_categorical=True, maximum_cardinality=True,
- minimum_cardinality=True, missing_target=True, encode_target=None)
-
+ minimum_cardinality=True, drop_duplicates=False, missing_target=True, encode_target=None)
+
Applies standard data cleaning steps on the dataset. Use the parameters
to choose which transformations to perform. The available steps are:
@@ -1518,7 +1518,7 @@ Data cleaning
method impute(strat_num="drop", strat_cat="drop", min_frac_rows=None, min_frac_cols=None, missing=None)
-
+
Impute or remove missing values according to the selected strategy. Also removes
rows and columns with too many missing values. The imputer is fitted only on the
training set to avoid data leakage. Use the missing attribute to customize what
@@ -1528,7 +1528,7 @@
Data cleaning
method encode(strategy="LeaveOneOut", max_onehot=10, frac_to_other=None)
-
+
Perform encoding of categorical features. The encoding type depends on the
number of unique values in the column:
@@ -1545,7 +1545,7 @@ Data cleaning
method prune(strategy="z-score", method="drop", max_sigma=3, include_target=False, **kwargs)
-
+
Prune outliers from the training set. The definition of outlier depends
on the selected strategy and can greatly differ from one each other.
Ignores categorical columns. Only outliers from the training set are pruned
@@ -1572,7 +1572,7 @@
Feature engineering
method feature_generation(strategy="DFS", n_features=None, generations=20, population=500, operators=None)
-
+
Use Deep feature Synthesis or a genetic algorithm to create new combinations
of existing features to capture the non-linear relations between the original
features. See FeatureGenerator for
@@ -1582,7 +1582,7 @@
Feature engineering
method feature_selection(strategy=None, solver=None, n_features=None,
max_frac_repeated=1., max_correlation=1., **kwargs)
-
+
Remove features according to the selected strategy. Ties between features with
equal scores are broken in an unspecified way. Also removes features with
too low variance and finds pairs of collinear features based on the Pearson
@@ -1635,21 +1635,21 @@
Training
method run(models, metric=None, greater_is_better=True, needs_proba=False, needs_threshold=False,
n_calls=10, n_initial_points=5, est_params=None, bo_params=None, bagging=0)
-
+
Runs a DirectRegressor instance.
method successive_halving(models, metric=None, greater_is_better=True, needs_proba=False,
needs_threshold=False, skip_runs=0, n_calls=0, n_initial_points=5,
est_params=None, bo_params=None, bagging=0)
-
+
Runs a SuccessiveHalvingRegressor instance.
method train_sizing(models, metric=None, greater_is_better=True, needs_proba=False,
needs_threshold=False, train_sizes=np.linspace(0.2, 1.0, 5), n_calls=0,
n_initial_points=5, est_params=None, bo_params=None, bagging=0)
-
+
Runs a TrainSizingRegressor instance.
Example
diff --git a/docs/API/data_cleaning/balancer/index.html b/docs/API/data_cleaning/balancer/index.html
index 9357c9a45..ed2840c1b 100644
--- a/docs/API/data_cleaning/balancer/index.html
+++ b/docs/API/data_cleaning/balancer/index.html
@@ -590,7 +590,7 @@ Balancer
class atom.data_cleaning.Balancer(strategy="ADASYN", n_jobs=1, verbose=0, logger=None, random_state=None, **kwargs)
-
+
Balance the number of samples per class in the target column. Use only for classification tasks.
This class can be accessed from atom through the
balance method. Read more in
@@ -727,7 +727,7 @@
Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -746,7 +746,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -784,7 +784,7 @@ Methods
method transform(X, y)
-
+
Oversample or undersample the data.
diff --git a/docs/API/data_cleaning/cleaner/index.html b/docs/API/data_cleaning/cleaner/index.html
index 2010bfdf8..7f6ef05e8 100644
--- a/docs/API/data_cleaning/cleaner/index.html
+++ b/docs/API/data_cleaning/cleaner/index.html
@@ -591,7 +591,7 @@ Cleaner
class atom.data_cleaning.Cleaner(prohibited_types=None, maximum_cardinality=True, minimum_cardinality=True,
strip_categorical=True, drop_duplicates=False, missing_target=True,
encode_target=True, verbose=0, logger=None)
-
+
Performs standard data cleaning steps on a dataset. Use the parameters to choose
which transformations to perform. The available steps are:
@@ -783,7 +783,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -802,7 +802,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -843,7 +843,7 @@ Methods
method transform(X, y=None)
-
+
Apply the data cleaning steps on the data.
diff --git a/docs/API/data_cleaning/encoder/index.html b/docs/API/data_cleaning/encoder/index.html
index 4f4753307..7dff221cc 100644
--- a/docs/API/data_cleaning/encoder/index.html
+++ b/docs/API/data_cleaning/encoder/index.html
@@ -586,7 +586,7 @@ Encoder
class atom.data_cleaning.Encoder(strategy="LeaveOneOut", max_onehot=10,
frac_to_other=None, verbose=0, logger=None, **kwargs)
-
+
Perform encoding of categorical features. The encoding type depends on
the number of classes in the column:
@@ -698,7 +698,7 @@ Methods
method fit(X, y)
-
+
Fit to data.
@@ -783,7 +783,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -802,7 +802,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -840,7 +840,7 @@ Methods
method transform(X, y=None)
-
+
Encode the data.
diff --git a/docs/API/data_cleaning/imputer/index.html b/docs/API/data_cleaning/imputer/index.html
index 6d1122bff..260271bc2 100644
--- a/docs/API/data_cleaning/imputer/index.html
+++ b/docs/API/data_cleaning/imputer/index.html
@@ -590,7 +590,7 @@ Imputer
class atom.data_cleaning.Imputer(strat_num="drop", strat_cat="drop", min_frac_rows=None,
min_frac_cols=None, verbose=0, logger=None)
-
+
Impute or remove missing values according to the selected strategy.
Also removes rows and columns with too many missing values. Use
the missing attribute to customize what are considered "missing
@@ -720,7 +720,7 @@
Methods
method fit(X, y=None)
-
+
Fit to data.
@@ -809,7 +809,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -828,7 +828,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -866,7 +866,7 @@ Methods
method transform(X, y=None)
-
+
Impute the data. Note that leaving y=None can lead to inconsistencies in
data length between X and y if rows are dropped during the transformation.
diff --git a/docs/API/data_cleaning/pruner/index.html b/docs/API/data_cleaning/pruner/index.html
index b557f2b56..7c2e9c3c9 100644
--- a/docs/API/data_cleaning/pruner/index.html
+++ b/docs/API/data_cleaning/pruner/index.html
@@ -590,7 +590,7 @@ Pruner
class atom.data_cleaning.Pruner(strategy="z-score", method="drop", max_sigma=3,
include_target=False, verbose=0, logger=None, **kwargs)
-
+
Replace or remove outliers. The definition of outlier depends
on the selected strategy and can greatly differ from one
another. Ignores categorical columns. This class can be accessed
@@ -736,7 +736,7 @@
Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -755,7 +755,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -793,7 +793,7 @@ Methods
method transform(X, y=None)
-
+
Apply the outlier strategy on the data.
diff --git a/docs/API/data_cleaning/scaler/index.html b/docs/API/data_cleaning/scaler/index.html
index 044b190fe..3ee3445a7 100644
--- a/docs/API/data_cleaning/scaler/index.html
+++ b/docs/API/data_cleaning/scaler/index.html
@@ -589,7 +589,7 @@
Scaler
class atom.data_cleaning.Scaler(strategy="standard", verbose=0, logger=None)
-
+
This class applies one of sklearn's scalers. It also returns a dataframe
when provided, and it ignores non-numerical columns (instead of raising
an exception). This class can be accessed from atom through the
@@ -692,7 +692,7 @@
Methods
method fit(X, y=None)
-
+
Compute the mean and std to be used for later scaling.
@@ -769,7 +769,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -788,7 +788,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
@@ -826,7 +826,7 @@ Methods
method transform(X, y=None)
-
+
Perform standardization by centering and scaling.
diff --git a/docs/API/feature_engineering/feature_generator/index.html b/docs/API/feature_engineering/feature_generator/index.html
index 8918a178b..b7c70810b 100644
--- a/docs/API/feature_engineering/feature_generator/index.html
+++ b/docs/API/feature_engineering/feature_generator/index.html
@@ -838,7 +838,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -857,7 +857,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
diff --git a/docs/API/feature_engineering/feature_selector/index.html b/docs/API/feature_engineering/feature_selector/index.html
index df7af46a7..d4bc2d485 100644
--- a/docs/API/feature_engineering/feature_selector/index.html
+++ b/docs/API/feature_engineering/feature_selector/index.html
@@ -950,7 +950,7 @@ Methods
method log(msg, level=0)
-
+
Write a message to the logger and print it to stdout.
@@ -969,19 +969,19 @@ Methods
method plot_pca(title=None, figsize=(10, 6), filename=None, display=True)
-
+
Plot the explained variance ratio vs the number of components.
See plot_pca for a description of the parameters.
method plot_components(show=None, title=None, figsize=None, filename=None, display=True)
-
+
Plot the explained variance ratio per components.
See plot_components for a description of the parameters.
method plot_rfecv(title=None, figsize=(10, 6), filename=None, display=True)
-
+
Plot the scores obtained by the estimator fitted on every subset of
the data. See plot_rfecv for a description of the parameters.
@@ -992,7 +992,7 @@ Methods
method save(filename=None)
-
+
Save the instance to a pickle file.
diff --git a/docs/API/models/adab/index.html b/docs/API/models/adab/index.html
index ca4c5506b..456f3245e 100644
--- a/docs/API/models/adab/index.html
+++ b/docs/API/models/adab/index.html
@@ -593,9 +593,9 @@
AdaBoost (AdaB)
AdaBoost is a meta-estimator that begins by fitting a classifier/regressor on
- the original dataset and then fits additional copies of the algorithm on the
- same dataset but where the weights of instances are adjusted according to the
- error of the current prediction.
+the original dataset and then fits additional copies of the algorithm on the
+same dataset but where the weights of instances are adjusted according to the
+error of the current prediction.
Corresponding estimators are:
- AdaBoostClassifier
@@ -608,9 +608,9 @@
AdaBoost (AdaB)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
algorithm parameter is only used with AdaBoostClassifier.
- The
loss parameter is only used with AdaBoostRegressor.
- The
random_state parameter is set equal to that of the trainer.
@@ -733,7 +733,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -745,8 +746,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -758,7 +759,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -768,9 +770,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -788,9 +790,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -842,8 +844,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.adab.plot_permutation_importance()
- or atom.adab.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.adab.plot_permutation_importance()
+or atom.adab.predict(X).The remaining utility methods can be found hereunder:
@@ -879,13 +881,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -902,13 +904,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -921,13 +924,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -936,7 +939,7 @@ Methods
metric: str or None, optional (default=None)
Name of the metric to calculate. Choose from any of sklearn's SCORERS
- or one of the following custom metrics (only if classifier):
+or one of the following custom metrics (only if classifier):
- "cm" for the confusion matrix.
- "tn" for true negatives.
@@ -970,7 +973,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/ard/index.html b/docs/API/models/ard/index.html
index 7ca1d6a75..1c3317100 100644
--- a/docs/API/models/ard/index.html
+++ b/docs/API/models/ard/index.html
@@ -593,9 +593,9 @@
Automatic Relevance Determination (ARD)
Automatic Relevance Determination is very similar to Bayesian Ridge, but
- can lead to sparser coefficients. Fit the weights of a regression model, using an
- ARD prior. The weights of the regression model are assumed to be in Gaussian
- distributions.
+can lead to sparser coefficients. Fit the weights of a regression model, using an
+ARD prior. The weights of the regression model are assumed to be in Gaussian
+distributions.
Corresponding estimators are:
- ARDRegression
@@ -606,9 +606,9 @@
Automatic Relevance Determination
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
@@ -732,7 +732,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -744,8 +745,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -757,7 +758,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -767,9 +769,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -787,9 +789,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -817,8 +819,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.ard.plot_permutation_importance()
- or atom.ard.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.ard.plot_permutation_importance()
+or atom.ard.predict(X).The remaining utility methods can be found hereunder:
@@ -849,13 +851,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -868,13 +871,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -905,7 +908,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/bag/index.html b/docs/API/models/bag/index.html
index f160c202f..3639527a7 100644
--- a/docs/API/models/bag/index.html
+++ b/docs/API/models/bag/index.html
@@ -592,12 +592,13 @@
Bagging (Bag)
-Bagging uses an ensemble meta-estimator that fits base classifiers/regressors each on
- random subsets of the original dataset and then aggregate their individual predictions
- (either by voting or by averaging) to form a final prediction. Such a meta-estimator
- can typically be used as a way to reduce the variance of a black-box estimator
- (e.g., a decision tree), by introducing randomization into its construction
- procedure and then making an ensemble out of it.
+Bagging uses an ensemble meta-estimator that fits base classifiers/regressors
+each on random subsets of the original dataset and then aggregate their
+individual predictions (either by voting or by averaging) to form a final
+prediction. Such a meta-estimator can typically be used as a way to reduce
+the variance of a black-box estimator (e.g., a decision tree),
+by introducing randomization into its construction procedure and then
+making an ensemble out of it.
Corresponding estimators are:
- BaggingClassifier
@@ -610,9 +611,9 @@
Bagging (Bag)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
n_jobs and random_state parameters are set equal to those of the
trainer.
@@ -738,7 +739,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -750,8 +752,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -763,7 +765,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -773,9 +776,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -793,9 +796,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -839,8 +842,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.bag.plot_permutation_importance()
- or atom.bag.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.bag.plot_permutation_importance()
+or atom.bag.predict(X).The remaining utility methods can be found hereunder:
@@ -876,13 +879,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -899,13 +902,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -918,13 +922,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -967,7 +971,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/bnb/index.html b/docs/API/models/bnb/index.html
index 927390d66..caae55859 100644
--- a/docs/API/models/bnb/index.html
+++ b/docs/API/models/bnb/index.html
@@ -592,10 +592,10 @@
Bernoulli Naive Bayes (BNB)
-Bernoulli Naive Bayes implements the Naive Bayes algorithm for multivariate Bernoulli
- models. Like Multinomial Naive bayes (MNB), this classifier is suitable for
- discrete data. The difference is that while MNB works with occurrence counts, BNB
- is designed for binary/boolean features.
+Bernoulli Naive Bayes implements the Naive Bayes algorithm for multivariate
+Bernoulli models. Like Multinomial Naive bayes (MNB), this
+classifier is suitable for discrete data. The difference is that while
+MNB works with occurrence counts, BNB is designed for binary/boolean features.
Corresponding estimators are:
- BernoulliNB
@@ -606,9 +606,9 @@
Bernoulli Naive Bayes (BNB)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
@@ -721,7 +721,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -733,8 +734,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -746,7 +747,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -756,9 +758,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -776,9 +778,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -822,8 +824,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.bnb.plot_permutation_importance() or atom.bnb.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.bnb.plot_permutation_importance() or atom.bnb.predict(X).
+The remaining utility methods can be found hereunder:
@@ -859,13 +861,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset.
| Parameters: |
@@ -882,13 +884,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -901,13 +904,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -950,7 +953,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/br/index.html b/docs/API/models/br/index.html
index b8f3f0126..e7040829a 100644
--- a/docs/API/models/br/index.html
+++ b/docs/API/models/br/index.html
@@ -592,9 +592,9 @@
Bayesian Ridge (BR)
-Bayesian regression techniques can be used to include regularization parameters in the
- estimation procedure: the regularization parameter is not set in a hard sense but
- tuned to the data at hand.
+Bayesian regression techniques can be used to include regularization
+parameters in the estimation procedure: the regularization parameter
+is not set in a hard sense but tuned to the data at hand.
Corresponding estimators are:
- BayesianRidge
@@ -605,9 +605,9 @@
Bayesian Ridge (BR)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
@@ -731,7 +731,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -743,8 +744,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -756,7 +757,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -766,9 +768,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -786,9 +788,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -816,8 +818,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.br.plot_permutation_importance()
- or atom.br.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.br.plot_permutation_importance()
+or atom.br.predict(X).The remaining utility methods can be found hereunder:
@@ -848,13 +850,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -867,13 +870,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -904,7 +907,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/catb/index.html b/docs/API/models/catb/index.html
index 6ed85d9e7..b106372f6 100644
--- a/docs/API/models/catb/index.html
+++ b/docs/API/models/catb/index.html
@@ -592,8 +592,8 @@
CatBoost (CatB)
-CatBoost is a machine learning method based on gradient boosting over decision trees.
- Main advantages of CatBoost:
+CatBoost is a machine learning method based on gradient boosting over
+decision trees. Main advantages of CatBoost:
- Superior quality when compared with other GBDT models on many datasets.
- Best in class prediction speed.
@@ -608,16 +608,16 @@ CatBoost (CatB)
Read more in CatBoost's documentation.
Note
-
CatBoost allows early stopping to stop
-the training of unpromising models prematurely!
+
CatBoost allows early stopping
+to stop the training of unpromising models prematurely!
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
bootstrap_type parameter is set to "Bernoulli" to allow for the subsample parameter.
- The
num_leaves and min_child_samples parameters are not available for the
CPU implementation.
@@ -750,7 +750,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -762,8 +763,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -775,8 +776,9 @@ Utility attributes
evals: dict
-Dictionary of the metric calculated during training. The metric is provided by the estimator's
- package and is different for every task. Available keys are:
+Dictionary of the metric calculated during training. The metric is
+provided by the estimator's package and is different for every task.
+Available keys are:
- "metric": Name of the metric.
- "train": List of scores calculated on the training set.
@@ -785,7 +787,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -795,9 +798,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -815,9 +818,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -861,8 +864,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.catb.plot_permutation_importance() or atom.catb.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.catb.plot_permutation_importance() or atom.catb.predict(X).
+The remaining utility methods can be found hereunder:
@@ -898,13 +901,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -921,13 +924,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -940,13 +944,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -989,7 +993,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/catnb/index.html b/docs/API/models/catnb/index.html
index 556dbcb31..88f5e3033 100644
--- a/docs/API/models/catnb/index.html
+++ b/docs/API/models/catnb/index.html
@@ -592,7 +592,8 @@
Categorical Naive Bayes (CatNB)
-Categorical Naive Bayes implements the Naive Bayes algorithm for categorical features.
+Categorical Naive Bayes implements the Naive Bayes algorithm for
+categorical features.
Corresponding estimators are:
- CategoricalNB
@@ -603,9 +604,9 @@
Categorical Naive Bayes (CatNB)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
@@ -718,7 +719,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -730,8 +732,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -743,7 +745,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -753,9 +756,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -773,9 +776,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -819,8 +822,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.catnb.plot_permutation_importance() or atom.catnb.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.catnb.plot_permutation_importance() or atom.catnb.predict(X).
+The remaining utility methods can be found hereunder:
@@ -856,13 +859,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset.
| Parameters: |
@@ -879,13 +882,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -898,13 +902,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -947,7 +951,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/cnb/index.html b/docs/API/models/cnb/index.html
index 6247a0298..dd9e6fee1 100644
--- a/docs/API/models/cnb/index.html
+++ b/docs/API/models/cnb/index.html
@@ -592,9 +592,9 @@
Complement Naive Bayes (CNB)
-The Complement Naive Bayes classifier was designed to correct the “severe assumptions”
- made by the standard Multinomial Naive Bayes classifier. It is particularly
- suited for imbalanced data sets.
+The Complement Naive Bayes classifier was designed to correct the
+“severe assumptions” made by the standard Multinomial Naive Bayes
+classifier. It is particularly suited for imbalanced data sets.
Corresponding estimators are:
- ComplementNB
@@ -605,9 +605,9 @@
Complement Naive Bayes (CNB)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
@@ -724,7 +724,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -736,8 +737,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -749,7 +750,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -759,9 +761,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -779,9 +781,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -825,8 +827,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.cnb.plot_permutation_importance() or atom.cnb.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.cnb.plot_permutation_importance() or atom.cnb.predict(X).
+The remaining utility methods can be found hereunder:
@@ -862,13 +864,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset.
| Parameters: |
@@ -885,13 +887,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -904,13 +907,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -953,7 +956,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/en/index.html b/docs/API/models/en/index.html
index 4a9245141..da13e11d9 100644
--- a/docs/API/models/en/index.html
+++ b/docs/API/models/en/index.html
@@ -603,9 +603,9 @@ Elastic Net (EN)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
random_state parameter is set equal to that of the trainer.
@@ -722,7 +722,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -734,8 +735,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -747,7 +748,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -757,9 +759,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -777,9 +779,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -807,8 +809,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.en.plot_permutation_importance()
- or atom.en.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.en.plot_permutation_importance()
+ or atom.en.predict(X).The remaining utility methods can be found hereunder:
@@ -839,13 +841,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -858,13 +861,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -895,7 +898,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/et/index.html b/docs/API/models/et/index.html
index 05aedf8c9..e2ef4337b 100644
--- a/docs/API/models/et/index.html
+++ b/docs/API/models/et/index.html
@@ -592,9 +592,10 @@
-Extra-Trees use a meta estimator that fits a number of randomized decision trees
- (a.k.a. extra-trees) on various sub-samples of the dataset and uses averaging
- to improve the predictive accuracy and control over-fitting.
+Extra-Trees use a meta estimator that fits a number of randomized
+decision trees (a.k.a. extra-trees) on various sub-samples of the
+dataset and uses averaging to improve the predictive accuracy and
+control over-fitting.
Corresponding estimators are:
- ExtraTreesClassifier
@@ -607,9 +608,9 @@
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
max_samples parameter is only used when bootstrap = True.
- The
n_jobs and random_state parameters are set equal to those of the
trainer.
@@ -757,7 +758,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -769,8 +771,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -782,7 +784,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -792,9 +795,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -812,9 +815,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -858,8 +861,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.et.plot_permutation_importance()
- or atom.et.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.et.plot_permutation_importance()
+or atom.et.predict(X).The remaining utility methods can be found hereunder:
@@ -895,13 +898,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -918,13 +921,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -937,13 +941,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -986,7 +990,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/gbm/index.html b/docs/API/models/gbm/index.html
index 95274c5e6..274e05d44 100644
--- a/docs/API/models/gbm/index.html
+++ b/docs/API/models/gbm/index.html
@@ -592,11 +592,12 @@
Gradient Boosting Machine (GBM)
-A Gradient Boosting Machine builds an additive model in a forward stage-wise
- fashion; it allows for the optimization of arbitrary differentiable loss
- functions. In each stage n_classes_ regression trees are fit on the negative
- gradient of the binomial or multinomial deviance loss function. Binary
- classification is a special case where only a single regression tree is induced.
+A Gradient Boosting Machine builds an additive model in a forward
+stage-wise fashion; it allows for the optimization of arbitrary
+differentiable loss functions. In each stage n_classes_ regression
+trees are fit on the negative gradient of the binomial or multinomial
+deviance loss function. Binary classification is a special case where
+only a single regression tree is induced.
Corresponding estimators are:
- GradientBoostingClassifier
@@ -609,9 +610,9 @@
Gradient Boosting Machine (GBM)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- For multiclass classification tasks, the
loss parameter is always set to "deviance".
- The
alpha parameter is only used when loss = "huber" or "quantile".
- The
random_state parameter is set equal to that of the trainer.
@@ -767,7 +768,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -779,8 +781,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -792,7 +794,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -802,9 +805,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -822,9 +825,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -876,8 +879,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.gbm.plot_permutation_importance()
- or atom.gbm.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.gbm.plot_permutation_importance()
+or atom.gbm.predict(X).The remaining utility methods can be found hereunder:
@@ -913,13 +916,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -936,13 +939,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -955,13 +959,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -1004,7 +1008,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/gnb/index.html b/docs/API/models/gnb/index.html
index ab62a2909..990581eaa 100644
--- a/docs/API/models/gnb/index.html
+++ b/docs/API/models/gnb/index.html
@@ -592,8 +592,9 @@
Gaussian Naive bayes (GNB)
-Gaussian Naive Bayes implements the Naive Bayes algorithm for classification. The
- likelihood of the features is assumed to be Gaussian.
+Gaussian Naive Bayes implements the Naive Bayes algorithm for
+classification. The likelihood of the features is assumed to
+be Gaussian.
Corresponding estimators are:
- GaussianNB
@@ -604,9 +605,9 @@
Gaussian Naive bayes (GNB)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- GNB has no parameters to tune with the BO.
@@ -694,8 +695,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -707,7 +708,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -736,9 +738,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -782,8 +784,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.gnb.plot_permutation_importance() or atom.gnb.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.gnb.plot_permutation_importance() or atom.gnb.predict(X).
+The remaining utility methods can be found hereunder:
@@ -819,13 +821,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset.
| Parameters: |
@@ -842,13 +844,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -861,13 +864,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -910,7 +913,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/gp/index.html b/docs/API/models/gp/index.html
index db726b97e..abdba62ff 100644
--- a/docs/API/models/gp/index.html
+++ b/docs/API/models/gp/index.html
@@ -592,19 +592,22 @@
Gaussian Process (GP)
-Gaussian Processes are a generic supervised learning method designed to solve
- regression and probabilistic classification problems. The advantages of Gaussian processes are:
+Gaussian Processes are a generic supervised learning method designed
+to solve regression and probabilistic classification problems. The
+advantages of Gaussian processes are:
- The prediction interpolates the observations.
-- The prediction is probabilistic (Gaussian) so that one can compute empirical confidence
- intervals and decide based on those if one should refit (online fitting, adaptive fitting)
- the prediction in some region of interest.
+- The prediction is probabilistic (Gaussian) so that one can compute
+ empirical confidence intervals and decide based on those if one
+ should refit (online fitting, adaptive fitting) the prediction in
+ some region of interest.
The disadvantages of Gaussian processes include:
-- They are not sparse, i.e. they use the whole samples/features information to perform the prediction.
-- They lose efficiency in high dimensional spaces, namely when the number of features
- exceeds a few dozens.
+- They are not sparse, i.e. they use the whole samples/features
+ information to perform the prediction.
+- They lose efficiency in high dimensional spaces, namely when the
+ number of features exceeds a few dozens.
Corresponding estimators are:
@@ -618,9 +621,9 @@ Gaussian Process (GP)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- GP has no parameters to tune with the BO.
@@ -707,8 +710,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -720,7 +723,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -749,9 +753,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -795,8 +799,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.gp.plot_permutation_importance() or atom.gp.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.gp.plot_permutation_importance() or atom.gp.predict(X).
+The remaining utility methods can be found hereunder:
@@ -832,13 +836,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -855,13 +859,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -874,13 +879,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -923,7 +928,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/knn/index.html b/docs/API/models/knn/index.html
index 65e057ecf..064d7c537 100644
--- a/docs/API/models/knn/index.html
+++ b/docs/API/models/knn/index.html
@@ -592,9 +592,10 @@
K-Nearest Neighbors (KNN)
-K-Nearest Neighbors, as the name clearly indicates, implements the k-nearest
- neighbors vote. For regression, the target is predicted by local interpolation
- of the targets associated of the nearest neighbors in the training set.
+K-Nearest Neighbors, as the name clearly indicates, implements the
+k-nearest neighbors vote. For regression, the target is predicted
+by local interpolation of the targets associated of the nearest
+neighbors in the training set.
Corresponding estimators are:
- KNeighborsClassifier
@@ -607,9 +608,9 @@
K-Nearest Neighbors (KNN)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
n_jobs parameter is set equal to that of the trainer.
@@ -734,7 +735,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -746,8 +748,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -759,7 +761,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -769,9 +772,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -789,9 +792,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -835,8 +838,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.knn.plot_permutation_importance()
- or atom.knn.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.knn.plot_permutation_importance()
+or atom.knn.predict(X).The remaining utility methods can be found hereunder:
@@ -872,13 +875,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -895,13 +898,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -914,13 +918,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -963,7 +967,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/ksvm/index.html b/docs/API/models/ksvm/index.html
index 85ed95996..48f373c6a 100644
--- a/docs/API/models/ksvm/index.html
+++ b/docs/API/models/ksvm/index.html
@@ -593,10 +593,10 @@
Kernel-SVM (kSVM)
The implementation of the Kernel (non-linear) Support Vector Machine is
- based on libsvm. The fit time scales at least quadratically with the number
- of samples and may be impractical beyond tens of thousands of samples. For
- large datasets consider using a Linear Support Vector Machine
- or a Stochastic Gradient descent model instead.
+based on libsvm. The fit time scales at least quadratically with the
+number of samples and may be impractical beyond tens of thousands of
+samples. For large datasets consider using a Linear Support Vector Machine
+or a Stochastic Gradient descent model instead.
The multiclass support is handled according to a one-vs-one scheme.
Corresponding estimators are:
@@ -610,9 +610,9 @@ Kernel-SVM (kSVM)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
degree parameter is only used when kernel = "poly".
- The
gamma parameter is always set to "scale" when kernel = "poly".
- The
coef0 parameter is only used when kernel = "rbf".
@@ -744,7 +744,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -756,8 +757,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -769,7 +770,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -779,9 +781,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -799,9 +801,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -837,8 +839,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the models, e.g. atom.ksvm.plot_permutation_importance() or atom.ksvm.predict(X).
- The remaining utility methods can be found hereunder:
+can be called directly from the models, e.g. atom.ksvm.plot_permutation_importance() or atom.ksvm.predict(X).
+The remaining utility methods can be found hereunder:
@@ -874,13 +876,13 @@ Methods
method calibrate(**kwargs)
-
-Applies probability calibration on the estimator. The calibration is done using the
- CalibratedClassifierCV
- class from sklearn. The calibrator is trained via cross-validation on a subset
- of the training data, using the rest to fit the calibrator. The new classifier will
- replace the estimator attribute. After calibrating, all prediction attributes will
- reset. Only if classifier.
+
+Applies probability calibration on the estimator. The calibration is done
+using the CalibratedClassifierCV
+class from sklearn. The calibrator is trained via cross-validation on a
+subset of the training data, using the rest to fit the calibrator. The new
+classifier will replace the estimator attribute. After calibrating, all
+prediction attributes will reset. Only if classifier.
| Parameters: |
@@ -897,13 +899,14 @@ Methods
method delete()
-
+
Delete the model from the trainer.
method rename(name=None)
-
-Change the model's tag. Note that the acronym always stays at the beginning of the model's name.
+
+Change the model's tag. Note that the acronym always stays at the
+beginning of the model's name.
| Parameters: |
@@ -916,13 +919,13 @@ Methods
method reset_predictions()
-
+
Clear all the prediction attributes.
- Use this method to free some memory before saving the model.
+Use this method to free some memory before saving the model.
method scoring(metric=None, dataset="test", **kwargs)
-
+
Get the scoring for a specific metric.
@@ -965,7 +968,7 @@ Methods
method save_estimator(filename=None)
-
+
Save the estimator to a pickle file.
diff --git a/docs/API/models/lasso/index.html b/docs/API/models/lasso/index.html
index 0145bc7bb..b8f553f4d 100644
--- a/docs/API/models/lasso/index.html
+++ b/docs/API/models/lasso/index.html
@@ -603,9 +603,9 @@ Lasso Regression (Lasso)
Hyperparameters
-- By default, the estimator adopts the default parameters provided by its package.
- See the user guide on how to
- customize them.
+- By default, the estimator adopts the default parameters provided by
+ its package. See the user guide
+ on how to customize them.
- The
random_state parameter is set equal to that of the trainer.
@@ -718,7 +718,8 @@ Utility attributes
estimator: class
-Estimator instance with the best combination of hyperparameters fitted on the complete training set.
+Estimator instance with the best combination of hyperparameters fitted
+on the complete training set.
time_bo: str
@@ -730,8 +731,8 @@ Utility attributes
time_fit: str
-Time it took to train the model on the complete training set and calculate the
- metric(s) on the test set.
+Time it took to train the model on the complete training set and
+calculate the metric(s) on the test set.
metric_train: float or list
@@ -743,7 +744,8 @@ Utility attributes
metric_bagging: list
-Bagging's results with shape=(bagging,) for single-metric runs and shape=(metric, bagging) for multi-metric runs.
+Bagging's results with shape=(bagging,) for single-metric runs and
+shape=(metric, bagging) for multi-metric runs.
mean_bagging: float or list
@@ -753,9 +755,9 @@ Utility attributes
Standard deviation of the bagging's results. List of values for multi-metric runs.
-results: pd.DataFrame
+results: pd.Series
-Dataframe of the training results with the model acronym as index. Columns can include:
+Series of the training results. Columns include:
- metric_bo: Best score achieved during the BO.
- time_bo: Time spent on the BO.
@@ -773,9 +775,9 @@ Utility attributes
Prediction attributes
-The prediction attributes are not calculated until the attribute is called for the
- first time. This mechanism avoids having to calculate attributes that are never
- used, saving time and memory.
+The prediction attributes are not calculated until the attribute is
+called for the first time. This mechanism avoids having to calculate
+attributes that are never used, saving time and memory.
@@ -803,8 +805,8 @@ Prediction attributes
Methods
The majority of the plots and prediction methods
- can be called directly from the model, e.g. atom.lasso.plot_permutation_importance()
- or atom.lasso.predict(X). The remaining utility methods can be found hereunder:
+can be called directly from the model, e.g. atom.lasso.plot_permutation_importance()
+or atom.lasso.predict(X).The remaining utility methods can be found hereunder:
@@ -835,13 +837,14 @@ Methods
method delete()
-