OPPapers.com

Cibil Project

Proposed Model Details
The model that we have tried to formulate tries to predict the CIBIL Score by asking a few questions and then calculating an internal score, which is a reflection of the CIBIL Score. The internal score is termed as discount score since that score further predicts as to how much discount must be offered to the customers. Thus, the proposed model calculates a discount score, which is directly proportional to the CIBIL Score. The discount percent given to the customers depends on the discount score that they score.
The discount score is not only a direct function of discount score but also inversely depended on the default probability. The default probability is a inverse function of CIBIL Score. The higher the CIBIL Score, the lesser the default probability. The CIBIL Score has been divided in different bands based on the default probability, which is calculated by the past data of NPAs. The same CIBIL bands have been used to divide the discount score in bands. The discount percent for different discount score bands are different and the exact pricing for different bands are explained further.
The figure to the left shows the relationship between discount percent, discount score, CIBIL score and default probability as discussed in the above paragraphs.
Structure of the project
The entire project can be divided in three major sections. Each section will be explained separately in the following sub-sections.
Data Analysis And Model building
The first section deals with various steps of data analysis and the steps that led to the model development. The data analysis and model building section can be broadly classified into three sections and they are explained below:
CIBIL Reports
The first step of model building was data analysis and that started with studying the CBIL Reports. We read approx 60 reports, studied and noted down the various details given in the report. The profile that we studied can be summarized as stated below:...