Week 4 – Modelling Preparation

10 Jul 2026 - Sneha Dharmeche - climate risk, data science, feature engineering, NHS ERIC

Project overview

This week continued the preparation of the NHS ERIC dataset for predictive modelling.

A key step this week was creating new features that better represent NHS estate characteristics. I prepped ‘total backlog maintenance cost’ variable by combining the low, moderate, significant and high-risk backlog maintenance costs into a single measure. This will be used as the target variable for predicting factors associated with NHS estate costs.

I also explored approaches for representing building age vulnerability. Initially, I created an age vulnerability score based on the proportion of buildings within older age categories. However, after discussing this with my supervisor, I identified that this approach may not fully capture differences between moderately old buildings and much older infrastructure.

I am now investigating alternative approaches, such as weighted age vulnerability scores, to better represent the impact of building age on estate vulnerability.

After my one-week break, I hope to move from data preparation into building and comparing predictive models.

Highlights of this week

I attended a TI planning meeting this week, where members discussed their ongoing projects, research progress and future plans. This provided an opportunity to understand how different projects are structured. The meeting was also a good opportunity to catch-up with supervisors and other members of the team. There was also had a great selection of snacks!