Skip to content
Mohadata Logo
All Case Studies

AI & Machine Learning

National Housing Association

Predictive Models for Operational Risk

~£9M

Savings in 18 Months

~60%

Ad-hoc Reporting Reduction

50+

Self-Service Users Enabled

The Challenge

A large UK regulated housing association had no data science capability. Decision-making on rent arrears, property disrepair, and operational resource allocation was reactive rather than predictive, costing millions annually. Reporting was ad-hoc, audit preparation was manual, and business users had no self-service access to analytics.

Our Solution

We established a Data Science Centre of Excellence with a multi-year capability roadmap and operating model. Core tooling was implemented on Azure Databricks and Power BI. Predictive and optimisation models were deployed addressing rent arrears, property disrepair risk, and operational automation — generating ~£9M in savings within 18 months.

How We Did It

Capability Roadmap & Operating Model

Capability Roadmap & Operating Model

Defined a multi-year data science roadmap covering platform selection, model development priorities, governance requirements, and team scaling plans. Modernised analytics delivery using Agile (Scrum and Kanban).

Platform Implementation

Platform Implementation

Deployed Azure Databricks for model training and orchestration, integrated with a governed Power BI portal for self-service analytics. Introduced automated data lineage and quality controls, reducing audit preparation effort by ~50%.

Predictive Model Deployment

Predictive Model Deployment

Built and deployed predictive models for rent arrears likelihood, property disrepair risk scoring, and operational resource optimisation. Models were validated against historical outcomes and deployed into production workflows.

Self-Service & Citizen Analytics

Self-Service & Citizen Analytics

Implemented enterprise self-service analytics via a governed Power BI portal — enabling 50+ business users and reducing ad-hoc reporting demand by ~60%. Developed citizen analytics capability by upskilling non-data employees.

For the first time we could predict which properties were at risk before issues escalated — the savings were immediate and dramatic.

Director of Operations

National Housing Association

Technologies Used

Azure DatabricksPower BIPythonScikit-learnPredictive ModellingRisk ScoringAgile / Scrum / KanbanData Quality Automation

Ready to achieve similar results?

Start a Conversation