Data-Driven Insights on Damp in English Housing: Surveying Practices, Remedial Measures, Stock Analysis, and Predictive Modelling

Damp in residential buildings presents significant challenges to housing quality, occupant health, and energy efficiency. This thesis adopts a mixed-methods approach, integrating qualitative (content and thematic analysis) and quantitative (Analytic Hierarchy Process (AHP), machine learning, and statistical analysis) methods to provide insights that could enhance damp management in English housing. To achieve this, the study uses real-world data from housing association inspection reports, surveyor comments, photographic evidence, a damp specialist questionnaire survey, and the Energy Performance Certificate (EPC) database. The research first examines current damp management practices, revealing a reactive approach that relies heavily on visual inspections and inconsistent diagnostic methods. Analysis of survey responses from damp specialists reveals differing views on the frequency of use, accessibility, and effectiveness of both diagnostic methods and remedial measures for damp. Findings also expose a gap between theoretical best practices and real-world damp management. AHP is used to investigate decision-making in damp diagnostics and remedial measures by providing a structured framework that reduces subjective bias. It integrates multiple criteria, such as cost, effectiveness, feasibility, and expertise availability, to support more informed and systematic decision-making. Results show some advanced diagnostic tools and remedial measures were deprioritised due to cost, feasibility constraints, or lack of specialist expertise. Machine learning clustering is applied to 1,655 damp homes, identifying three distinct damp home profiles based on building characteristics and energy efficiency. A defect analysis highlights condensation as the most prevalent issue, primarily affecting bathrooms, bedrooms, and kitchens. Finally, a predictive machine learning model is developed to predict damp risk in over 35,000 homes. Random Forest performed best, with SHAP (SHapley Additive exPlanations) analysis identifying heating cost, energy consumption, wall efficiency, and construction age as key predictors. A Shiny app prototype demonstrates the feasibility of single-property damp risk assessment, though generalisability remains a challenge.