There are over 2 billion people in the world who don't have access to finance. Not only is this a massive business opportunity, currently over $300 billion and growing 20% annually, it has the potential to fundamentally change economic opportunity for the poor.
While there are 15,000 microfinance companies currently fighting for this market, they all have one common challenge, creating a more effective field staff to get away from branch based banking.
In microfinance, "credit officers" are the sales people, loan officers, and a collectors. They have to travel hours to remote villages in the developing world, source potential customers, convince them, and make finance decisions about people with no credit history. They do this all in paper records that they have to enter it into an outdated information management system, usually built in house, back at the office.
Three things need to change: 1) Make loan officers more efficient at prioritizing their sales efforts. 2) Provide loan officers with machine learning based risk scoring to help them make good finance decisions in real time. 3) Introduce accountability.
We decided to build a mobile app and machine learning credit model to link the sales pipeline with the credit decision process. All information collection has been brought into the flow and there is a product selection aid, that dynamically weights which products to offer based on the lead quality. We then utilized a random forest model that is automatically trained on any current Loans in the organization's Force.com account and robustly surface the predictive model through Heroku. The credit scoring model approves or denies a loan in real time. For the first time ever, a credit officer can take a deal from a lead to a signed document all in the field, allowing almost anyone to effectively become a loan officer in the developing world.
Log in or sign up for Devpost to join the conversation.