Recommendation Engine for banks using Artificial Neural Networks
The code tries to solve Santander Bank's Kaggle challenge for product Recommendation Engine. The independent variable for the engine are - Month_status_date, Customer_ID, Customer_Name, Employee_Index, Customer_country, Sex, Age, Join_date, New_customer, Relationship_Months, Relationship_flag, Last_date_Primary_Customer, Customer_type_begin_Month, Cust_Relation_type_begin_month, Residence_flag, Forigner_flag, Emp_spouse_flag, Channel_when_joined, Deceased_flag Address_type, Customer_address, Address_detail, Activity_flag, Gross_household_income, Segment. Using these features, the model provides a list of new products to be recommended to customers.
Code Requirements - Scikit-Learn, Tensorflow, Keras, Xgboost
The code also contains EDA and statistical inferences drawn from the features. Various algorithmic models are made and compared for best resulst.