MatchiFi

Smarter funding, for every business

Comment

Inspiration

MatchiFi is a digital matchmaking platform designed to close the persistent financing gap faced by small, medium, and micro enterprises (SMMEs) in Botswana. The cautious approach of commercial banks driven by credit rationing behaviours, lack of collateral from borrowers, poor credit histories, and perceived business risk leaves a large fraction of consumers unbanked. MatchiFi seeks to serve this community by availing information about funding products and conducting an analysis of financial information to give out a probable response to a funding application.

What it does

The application conducts quick financial analysis by automatically calculating key financial ratios. It takes in data from users and carries out the calculations, with minimal human intervention. The program also has a prescriptive model which uses the aforementioned ratios as inputs. The ratios are fed into a classification machine learning model and returns the outcome. We treated a loan application process as a classification problem with 2 outcomes: Reject or Approve. In future, the program will be able to return a list of appropriate funding options for the user based on their financial position.

How we built it

We used Figma and bravo to design the UI. The UI prompts the user to enter figures for the relevant variables. This data is then stored in an SQL database from which it will be sourced for analysis. The team also used python to develop the backend. For the matching algorithm, we employed a logistic regression model to analyze the financial data and classify the user accordingly. Additionally, we have a financial ratio calculator which yields the inputs needed for the ML model. To test the program, the team used Command Line Interface (CLI).

Challenges we ran into

We faced a lack of actual data to train the model used by the program. Therefore, used synthetic data for training and testing. Automating the program was a complex task which required high level application of APIs.

Accomplishments that we're proud of

Coordinating the different skillset of the team members to create a working prototype which embraces simplicity and ease of use. The team got to appreciate the limitations of the systems currently in place. We also produced an efficient financial analysis program to carry out the calculation of business metrics for business owners and lending institutions alike.

What we learned

Goal-oriented collaboration and synthesis of ideas. We also learned about the development of a data analysis pipeline as well as the integration of databases with algorithms.

What's next for MatchiFi

The team plans to have both web and mobile apps for MatchiFi fully developed and operational.

Built With

Share this project:

Updates