Capstone project

Installation

Run pip install -r requirements.txt to intall libraries used in the project.

Notebooks

• Capstone-Proposal.ipynb - is the notebook where capstone proposal is written.
• Base_Line.ipynb - logistic regression and random guess applied to the data set.
• Data_Exploration.ipynb - data exploration and visualizatin notebook
• Data_Merging.ipynb - merging of all files into single file data set
• isolation_forest_outliers.npy - isolation forest applied to data set to detect possible outliers
• Lightgbm.ipynb - Lightgbm algorithm application

How to run and get results

Data_Exploration.ipynb

Use to analyze data, see visulizations, anomalies, etc.

Base_Line.ipynb

To build base line it's needed to extract features first, for that there is Feature_Extraction.ipynb, it pre-process data using basics techniques and save tran and test joblibs. After that it's possible to build base line.

isolation_forest_outliers.npy

This is optional step to find outliers using isolation forest algorithm.

Data_Merging.ipynb

Run this notebook to merge files into single train and test csv's. This step is neccessary to run Lightgbm notebook.

Lightgbm.ipynb

Run this notebook to train and get results using Lightgbm algorithm. It saves submission for kaggle at the end.

Data set

Data set is hosted on kaggle platform. To download it follow this link and accept 'Terms and Conditions' of the project and use download button or kaggle command line tool to download files:
kaggle competitions download -c home-credit-default-risk

Submission

kaggle competitions submit -c home-credit-default-risk -f submission.csv -m "Message"