Kolluri-Vechicle_Insurance_Classifier with Imbalanced Dataset

Dataset:

Information about Dataset:

Demographics (gender, age, region code type),

Vehicles (Vehicle Age, Damage),

Policy (Premium, sourcing channel) etc.

Attributes

id -> unique identity (int)

Gender -> Male/Female (object-string)

Age -> Age (int)

Driving_License -> 0/1 (binary)

Region_Code -> (int)

Previously_Insured -> 0/1 (binary)

Vehicle_Age -> (object)

Vehicle_Damage -> Yes/No (binary)

Annual_Premium -> (int)

Policy_Sales_Channel -> (int)

Vintage -> (int)

Response -> 0/1 (binary) (response)

Task

Predict whether customer is interested in vehicle insurance based on previous data.

Oversampling and Undersampling

• Smote (Synthetic Minority Oversampling Technique)-(OverSampling).
• NearMiss (UnderSampling).

Performed Models

• Model-1 : Logistic Regression.
• Model-2 : RandomForest.
• Model-3 : Gradient Boosting.
• Hyperparameter Tuning