AutoTabular

AutoTabular automates machine learning tasks enabling you to easily achieve strong predictive performance in your applications. With just a few lines of code, you can train and deploy high-accuracy machine learning and deep learning models tabular data.

[Toc]

What's good in it?

• It is using the RAPIDS as back-end support, gives you the ability to execute end-to-end data science and analytics pipelines entirely on GPUs.
• It Supports many anomaly detection models: ,
• It using meta learning to accelerate model selection and parameter tuning.
• It is using many Deep Learning models for tabular data: Wide&Deep, DCN(Deep & Cross Network), FM, DeepFM, PNN ...
• It is using many machine learning algorithms: Baseline, Linear, Random Forest, Extra Trees, LightGBM, Xgboost, CatBoost, and Nearest Neighbors.
• It can compute Ensemble based on greedy algorithm from Caruana paper.
• It can stack models to build level 2 ensemble (available in Compete mode or after setting stack_models parameter).
• It can do features preprocessing, like: missing values imputation and converting categoricals. What is more, it can also handle target values preprocessing.
• It can do advanced features engineering, like: Golden Features, Features Selection, Text and Time Transformations.
• It can tune hyper-parameters with not-so-random-search algorithm (random-search over defined set of values) and hill climbing to fine-tune final models.

Installation

The sources for AutoTabular can be downloaded from the Github repo.

You can either clone the public repository:

# clone project
git clone https://apulis-gitlab.apulis.cn/apulis/AutoTabular/autotabular.git
# First, install dependencies
pip install -r requirements.txt

Once you have a copy of the source, you can install it with:

python setup.py install

Example

Next, navigate to any file and run it.

# module folder
cd example

# run module (example: mnist as your main contribution)
python binary_classifier_Titanic.py

Auto Feature generate & Selection

TODO

Deep Feature Synthesis

import featuretools as ft
import pandas as pd
from sklearn.datasets import load_iris

# Load data and put into dataframe
iris = load_iris()
df = pd.DataFrame(iris.data, columns=iris.feature_names)
df['species'] = iris.target
df['species'] = df['species'].map({
    0: 'setosa',
    1: 'versicolor',
    2: 'virginica'
})
# Make an entityset and add the entity
es = ft.EntitySet()
es.add_dataframe(
    dataframe_name='data', dataframe=df, make_index=True, index='index')
# Run deep feature synthesis with transformation primitives
feature_matrix, feature_defs = ft.dfs(
    entityset=es,
    max_depth=3,
    target_dataframe_name='data',
    agg_primitives=['mode', 'mean', 'max', 'count'],
    trans_primitives=[
        'add_numeric', 'multiply_numeric', 'cum_min', 'cum_mean', 'cum_max'
    ],
    groupby_trans_primitives=['cum_sum'])

print(feature_defs)
print(feature_matrix.head())
print(feature_matrix.ww)

GBDT Feature Generate

from autofe.feature_engineering.gbdt_feature import CatboostFeatureTransformer, GBDTFeatureTransformer, LightGBMFeatureTransformer, XGBoostFeatureTransformer

titanic = pd.read_csv('autotabular/datasets/data/Titanic.csv')
# 'Embarked' is stored as letters, so fit a label encoder to the train set to use in the loop
embarked_encoder = LabelEncoder()
embarked_encoder.fit(titanic['Embarked'].fillna('Null'))
# Record anyone travelling alone
titanic['Alone'] = (titanic['SibSp'] == 0) & (titanic['Parch'] == 0)
# Transform 'Embarked'
titanic['Embarked'].fillna('Null', inplace=True)
titanic['Embarked'] = embarked_encoder.transform(titanic['Embarked'])
# Transform 'Sex'
titanic.loc[titanic['Sex'] == 'female', 'Sex'] = 0
titanic.loc[titanic['Sex'] == 'male', 'Sex'] = 1
titanic['Sex'] = titanic['Sex'].astype('int8')
# Drop features that seem unusable. Save passenger ids if test
titanic.drop(['Name', 'Ticket', 'Cabin'], axis=1, inplace=True)

trainMeans = titanic.groupby(['Pclass', 'Sex'])['Age'].mean()

def f(x):
    if not np.isnan(x['Age']):  # not NaN
        return x['Age']
    return trainMeans[x['Pclass'], x['Sex']]

titanic['Age'] = titanic.apply(f, axis=1)
rows = titanic.shape[0]
n_train = int(rows * 0.77)
train_data = titanic[:n_train, :]
test_data = titanic[n_train:, :]

X_train = titanic.drop(['Survived'], axis=1)
y_train = titanic['Survived']

clf = XGBoostFeatureTransformer(task='classification')
clf.fit(X_train, y_train)
result = clf.concate_transform(X_train)
print(result)

clf = LightGBMFeatureTransformer(task='classification')
clf.fit(X_train, y_train)
result = clf.concate_transform(X_train)
print(result)

clf = GBDTFeatureTransformer(task='classification')
clf.fit(X_train, y_train)
result = clf.concate_transform(X_train)
print(result)

clf = CatboostFeatureTransformer(task='classification')
clf.fit(X_train, y_train)
result = clf.concate_transform(X_train)
print(result)

from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import roc_auc_score

lr = LogisticRegression()
x_train_gb, x_test_gb, y_train_gb, y_test_gb = train_test_split(
    result, y_train)
x_train, x_test, y_train, y_test = train_test_split(X_train, y_train)

lr.fit(x_train, y_train)
score = roc_auc_score(y_test, lr.predict(x_test))
print('LR with GBDT apply data, train data shape : {0}  auc: {1}'.format(
    x_train.shape, score))

lr = LogisticRegression()
lr.fit(x_train_gb, y_train_gb)
score = roc_auc_score(y_test_gb, lr.predict(x_test_gb))
print('LR with GBDT apply data, train data shape : {0}  auc: {1}'.format(
    x_train_gb.shape, score))

Golden Feature Generate

from autofe import GoldenFeatureTransform

titanic = pd.read_csv('autotabular/datasets/data/Titanic.csv')
embarked_encoder = LabelEncoder()
embarked_encoder.fit(titanic['Embarked'].fillna('Null'))
# Record anyone travelling alone
titanic['Alone'] = (titanic['SibSp'] == 0) & (titanic['Parch'] == 0)
# Transform 'Embarked'
titanic['Embarked'].fillna('Null', inplace=True)
titanic['Embarked'] = embarked_encoder.transform(titanic['Embarked'])
# Transform 'Sex'
titanic.loc[titanic['Sex'] == 'female', 'Sex'] = 0
titanic.loc[titanic['Sex'] == 'male', 'Sex'] = 1
titanic['Sex'] = titanic['Sex'].astype('int8')
# Drop features that seem unusable. Save passenger ids if test
titanic.drop(['Name', 'Ticket', 'Cabin'], axis=1, inplace=True)

trainMeans = titanic.groupby(['Pclass', 'Sex'])['Age'].mean()

def f(x):
    if not np.isnan(x['Age']):  # not NaN
        return x['Age']
    return trainMeans[x['Pclass'], x['Sex']]

titanic['Age'] = titanic.apply(f, axis=1)

X_train = titanic.drop(['Survived'], axis=1)
y_train = titanic['Survived']
print(X_train)
gbdt_model = GoldenFeatureTransform(
    results_path='./', ml_task='BINARY_CLASSIFICATION')
gbdt_model.fit(X_train, y_train)
results = gbdt_model.transform(X_train)
print(results)

Neural Network Embeddings

# data url
"""https://www.kaggle.com/c/house-prices-advanced-regression-techniques."""
data_dir = '/media/robin/DATA/datatsets/structure_data/house_price/train.csv'
data = pd.read_csv(
    data_dir,
    usecols=[
        'SalePrice', 'MSSubClass', 'MSZoning', 'LotFrontage', 'LotArea',
        'Street', 'YearBuilt', 'LotShape', '1stFlrSF', '2ndFlrSF'
    ]).dropna()

categorical_features = [
    'MSSubClass', 'MSZoning', 'Street', 'LotShape', 'YearBuilt'
]
output_feature = 'SalePrice'
label_encoders = {}
for cat_col in categorical_features:
    label_encoders[cat_col] = LabelEncoder()
    data[cat_col] = label_encoders[cat_col].fit_transform(data[cat_col])

dataset = TabularDataset(
    data=data, cat_cols=categorical_features, output_col=output_feature)

batchsize = 64
dataloader = DataLoader(dataset, batchsize, shuffle=True, num_workers=1)

cat_dims = [int(data[col].nunique()) for col in categorical_features]
emb_dims = [(x, min(50, (x + 1) // 2)) for x in cat_dims]
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = FeedForwardNN(
    emb_dims,
    no_of_cont=4,
    lin_layer_sizes=[50, 100],
    output_size=1,
    emb_dropout=0.04,
    lin_layer_dropouts=[0.001, 0.01]).to(device)
print(model)
num_epochs = 100
criterion = nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.1)
for epoch in range(num_epochs):
    for y, cont_x, cat_x in dataloader:
        cat_x = cat_x.to(device)
        cont_x = cont_x.to(device)
        y = y.to(device)
        # Forward Pass
        preds = model(cont_x, cat_x)
        loss = criterion(preds, y)
        # Backward Pass and Optimization
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
    print('loss:', loss)

License

This library is licensed under the Apache 2.0 License.

Contributing to AutoTabular

We are actively accepting code contributions to the AutoTabular project. If you are interested in contributing to AutoTabular, please contact me.