BI connector and demonstration

Fineract is largely used as a system of record for lending in many institutions globally. As such, most of the analysis is about portfolio health — how well are people paying back, and what are the key risks around repayments. This includes PAR, loan loss provisioning, write-offs, and trend lines for the same, along with basic stats like number of customers, the loan cycle they are on, aging of customer cycles, exceptions tracking, and correlation to any characteristic in the data. Currently Fineract does not have a analytics pipeline where vendors, users, institution can see these analytics. This project solves this problem. My plan is to build a complete ELT analytics platform. A python extractor will pull data from read replica of Fineract DB incrementally using a read only database user. It will push the data the PostgreSQL analytics warehouse. From there a dbt project will handle all the data transformation (delinquency calculations, PAR ratios, repayment behavior, portfolio health). I will show interactive dashboards with different type of charts, graphs, stats and analytics using Apache Superset. By the end of GSoC, the deliverables are: a working Python ELT extractor with both backfill and incremental modes, a fully tested 4-layer dbt warehouse model covering the Fineract lending domain, five production-oriented Superset dashboards (Portfolio Health, Delinquency and PAR, Repayment Behavior, Risk and Finance, Customer Lifecycle), Docker Compose setup so anyone can run the full stack locally, and complete documentation including a runbook and data dictionary. The whole stack is Apache 2.0 licensed, self-hostable, and designed so future contributors can add new analytics modules without changing what's already there.