Laying the data foundation and predictive modelling framework for a housing financing company operating primarily in the un-served, unreached and under-served markets in India


In the second decade of its existence, this company is decisively progressing onto a technology and business transformation. On one hand it is navigating the challenges of the overall economy, geo-political unrest, inflation, and pandemic. On the other hand, it is also using this time to drive improvements and efficiencies in the areas of customer experience, distribution network, credit growth and cyber-security.


Dhurin is supporting this journey on both the data and modelling levels.

1. Creation of Customer data layer

  • Multiple dashboards serving following goals:
  • Employee productivity assessment

2. Portfolio tracking

3. Machine Learning based scorecards needed at different stages of customer life cycle

Exhibit 1

Customer 360 degree view is the core task of every lending business. The view delivers value only when it is seamlessly available. Every lending institution spends a lot of money in procuring credit bureau data, but a lot of times lies highly underutilised due to this lack of “seamlessness”. We designed and developed a MySQL analytical database for descriptive and predictive analytics. This database feeds on regular credit bureau files and creates a bureau 360-degree view. Accurate extraction from the non-digitized, binary format bureau files; and appropriate table updating brought in accuracy, efficiency, and completeness in building this foundation data.

Exhibit 2

Employee productivity is a critical element. Assessing it and tracking it regularly are the regular activities in process improvements, be it the process of underwriting, the sales process or collections process. We collaborated in the process of defining the key performance indicators (KPIs or Metrics), led the process of design, development and deployment of effective dashboards for two major workforce of the company, namely, the credit underwriting team and the sales team. Platform used was Tableau. Likewise, in a similar process, dashboards were built for portfolio tracking.

Exhibit 3

Under-served and hard to reach market comes with its own share of data challenges. In the data-driven solution strategy, often ML based scorecard plays an important role. We have been continuously building solution strategies for the client touching various stages of the customer lifecycle, i.e. from acquisition to account management to collections. One noteworthy set of solutions is the collections strategy we build. In it, have built ML scorecards covering different stages of collections (like Bounce scorecards, scorecards predicting an account going from Bad-stage to Worse-stage).

The results

The lending operations of the client have always been strong. However, for non-linear expansion, the strategy needed strong components of both descriptive data insights and predictive (ML) scorecards. We are providing them both, with encouraging results.