Loss prediction model

• 

  Correlation

https://prezi.com/view/3AJ8ogdZVninYtseSOEw/

We are using Random Forest Classification to predict whether we have loss or not and then if we have loss, we will use Random Forest Regression to obtain expected loss.

Due to imbalance classes for the classification, we tried balanced random forest, SMOTE, weight random forest.

We have tried several regression model such as linear regression, support vector regression, decision tree, neural network.