Non-Recovery Prediction Models
Iterating on ground motion variables for a predictive model of non-recovery of homes in Nepal from the 2015 earthquake.
Overview
This research develops and refines predictive models to identify households that are at risk of not recovering after major earthquake events. Using data from the 2015 Nepal earthquake, we examine which ground motion variables and socioeconomic factors best predict long-term non-recovery.
Research Questions
- Which ground motion variables are most predictive of household non-recovery?
- How can predictive models support rapid response and resource allocation?
- What socioeconomic factors interact with physical damage to affect recovery outcomes?
Methodology
This project combines statistical analysis of ground motion data with household survey data to develop and validate predictive models. We use machine learning techniques to identify the most important predictors of non-recovery and assess model performance across different contexts.
Current Status
This research is ongoing. We are currently iterating on model variables and validating results against updated recovery data.