Captstone Project by Nishant Sharma

This is a comprehensive comparison of risk prediction models.

Here is what you need to see with respect to the steps followed->

1.Data analysis and visualization can be found in Xploratory_analysis.ipynb

2.Preprocessing and Feed Forward Deep Neural network training and scores can be found in deep_default.ipynb

3.Preprocessing,ensemble methods training and model evaluations can be found in Preprocessing_modeling_refinements_evaluations.ipynb

Results- Finally,Gradient Boosting with tuned parameters was selected with optimal training/testing scores accross 10 folds of data.

References:

http://ufldl.stanford.edu/tutorial/supervised/MultiLayerNeuralNetworks/

https://medium.com/towards-data-science/decision-trees-and-random-forests-df0c3123f991

http://www.fon.hum.uva.nl/praat/manual/Feedforward_neural_networks_1_1__The_learning_phase.html

http://vinhkhuc.github.io/2015/03/01/how-many-folds-for-cross-validation.html

http://fastml.com/classifier-calibration-with-platts-scaling-and-isotonic-regression/

http://scikit-learn.org/0.15/auto_examples/grid_search_digits.html

https://svds.com/learning-imbalanced-classes/

https://stats.stackexchange.com/questions/117643/why-use-stratified-cross-validation-why-does-this-not-damage-variance-related-b

https://keras.io/optimizers/

https://keras.io/models/sequential/

http://seaborn.pydata.org/generated/seaborn.boxplot.html#seaborn.boxplot

https://jmetzen.github.io/2015-04-14/calibration.html