"Proposal of a credit scoring model to one of the leaders of microfinance in Cameroon." By J. Tejionang

​

in order to improve the service of her clientele "Salariée", the financial institute appealed to us to poposer a model of scoring which would reduce considerably the time of decision of credit for this customer. 

 

Goals:

- Customer satisfaction (decision time reduced to  few minutes)

- Quantification of the risk.

- Best allocation of funds

- Reduction of human bias

- Ability to adjust the acceptance threshold to follow the changes in credit policy.

 

The financial institute provided the data about credtis granted over a certain period of time and monitored for at least 12 months. 

​

Using the Bale definition of the default (90 days or more in defaults) as a variable to explain, we had at our disposal 17 other Variables to explain it.

​

70% of the data was used to train the model and the remaining 30% to validate it.. 

​

Results:

- a logistic model was obtained with a very good discriminatory power (AUC = 75.4%)