Introduction

This repo contains scripts for competing in the Kaggle Santanter Customer Satisfaction competition.

Starting

The Jupyter Notebook data cleaning.ipynb is the starting point to clean up the data. Run this first to prep the data for the other scripts.

I tried out the following algorithms in the various scripts:

I ran these models on a few machines and had the results saved to a cloud MongoDB so I could find the best hyperparameters.

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
.gitignore		.gitignore
57053e51de3a9c0b13b3376a.csv		57053e51de3a9c0b13b3376a.csv
57053e51de3a9c0b13b3376a_calibrated.csv		57053e51de3a9c0b13b3376a_calibrated.csv
57053e51de3a9c0b13b3376a_isotonic_calibrated.csv		57053e51de3a9c0b13b3376a_isotonic_calibrated.csv
Analyze GBC Scores.ipynb		Analyze GBC Scores.ipynb
Fit GradingBoostingClassifier.ipynb		Fit GradingBoostingClassifier.ipynb
Hyperopt GradientBoostingClassifier.ipynb		Hyperopt GradientBoostingClassifier.ipynb
README.md		README.md
SVM exploration.ipynb		SVM exploration.ipynb
Santander Customer Satisfaction.ipynb		Santander Customer Satisfaction.ipynb
SelectPercentile Sandbox.ipynb		SelectPercentile Sandbox.ipynb
TPOT.ipynb		TPOT.ipynb
XGBoost.ipynb		XGBoost.ipynb
data cleaning.ipynb		data cleaning.ipynb
empty_submission.df		empty_submission.df
run_gbc_feature_selection.py		run_gbc_feature_selection.py
run_gbc_fs_2.py		run_gbc_fs_2.py
run_gradientboostingclassifier.py		run_gradientboostingclassifier.py
run_tpot.py		run_tpot.py
run_xgb_feature_selection.py		run_xgb_feature_selection.py
run_xgboost.py		run_xgboost.py
sample_submission.csv		sample_submission.csv
sandbox.ipynb		sandbox.ipynb
test.csv		test.csv
train.csv		train.csv