Accident level prediction based on categorical features
Most of the data is categorical so I tried to use decision tress and random forests. Tried to do some feature engineering. Tried different variations in feature engineering like : one hot encoding, removing features based on variance...
I have 10 -15 variantions of these files with different hyperparameters tuning.
Final accuracy i could get is around 92%