Inspiration

Loan triaging is manual and time-intensive, not real time based on customer details input . Literally the idea was to show a layman can develop AI model without using python and deploy to the web.

What it does

loans in terms of economic pay outs is a very important side of banking business system. many loan applications are supported by bound inputs to validate the eligibility for loan. Here our use-case is that, we would like to automate the loan eligibility away from gender bias.

How I built it

I built the solution using the sample customer loan and demographics datasets of Finastra portal. Then i used Peltarion to build an AI model . from the data, I trained the model with customer data minimizing gender bias and targeted 'loan status' at Peltarion and hosted the API at azure. built the React app to do the Make the loan form . the dependent / target variable is the Loan Status and we needed to develop a model using the rest of the dataset features to predict the target. In the model training and evaluation we drop the 'Loan status' and placed it as target variable.
To start the loan application

1) Do a Post JSON to the finastra Auth API With access token returned, create an analysis of source application loan form with a Post Json to the azure api.

Challenges I ran into

Displaying Ai prediction on website. find hosting on the internet, mostly
wish i could have begin earlier vs a week ago.
With the customer_loans dataset missing gender feature, i included it from customer demographics dataset

Accomplishments that I'm proud of

without previous knowledge of ML or using python able to Build a loan application form to Ai workflow, that calls REST AI api .

What I learned

I learned to build AI model with peltarion platform. Learned calling of finastra api
With the customer_loans dataset missing gender feature, i included it from customer demographics dataset

What's next for Loan Triage

Build robust loan application and Integrate with other software.

Share this project:

Updates