Kaggle Competition. Built a Wide & Deep model that predicts the probability that a driver will initiate an auto insurance claim in the next year (Binary Classification).
1_Data_pre_processing.ipynb contains the initial data analysis 2_wide_n_deep_classifier.py is the submitted classifier
Bokeh plots of pre_processing are in Pre_processing_plots folder (Github strips javascript plots from jupyter notebook. So, they are added seperately)
Other files are either data generation files (last two words of the file will be data_gen.py) or other classifiers I tried.
wide_and_deep (submitted)
RBMs with Logistic Regression
RBMs with SVMs
SMOTE (with vanilla NN and CNNs)
Undersampling (with vanilla NN and CNNs)
Auto encoders