research

Positive-unlabelled machine learning flags recorded damp and mould risk across 65,214 UK social housing dwellings

An open-access UK study applies positive-unlabelled machine learning to 65,214 social housing dwellings to prioritise damp and mould risk where most homes lack recorded complaints, achieving PU-AUC 0.776 with heating/fuel systems and household context as key predictors - a method landlords and regulators could use to target inspections and repairs.

Published 2026-10-29

Source publication date: 2026-10-22

Location

London, England, GB

Application status

unverified

An open-access UK study applies positive-unlabelled machine learning to 65,214 social housing dwellings to prioritise damp and mould risk where most homes lack recorded complaints, achieving PU-AUC 0.776 with heating/fuel systems and household context as key predictors - a method landlords and regulators could use to target inspections and repairs.

Read the original source for the full report and any eligibility rules, dates or health claims.

Record updated 2026-10-29. See corrections policy.