FFRS Analysis

Fire Victim Survivability Analysis Report

1. Executive Summary (Lay Interpretation)

This study analyzed over 5,000 fire incident records to identify the factors most strongly associated with victim survival. Our findings highlight three critical 'Life Safety' pillars:

1. Visibility is the strongest predictor of life: Victims found in zero-visibility conditions are roughly 90% less likely to survive than those found in clear conditions.
2. Smoke Alarms save lives: The presence of a working smoke alarm more than doubles the odds of survival.
3. Speed of Search matters: Initiating a search before or during fire knockdown significantly improves the chances of a victim being found alive.

2. Statistical Analysis & Findings

A. Predictor Importance (Random Forest)

The Random Forest model ranks variables based on how much 'information' they provide in predicting the outcome. The location of the victim and the room they were found in are the top indicators of the eventual outcome.

B. Odds of Survival (Logistic Regression)

Logistic regression allows us to quantify the 'Risk' or 'Benefit' of specific conditions. An Odds Ratio (OR) greater than 1 represents a protective factor, while an OR less than 1 represents an increased risk.

C. General Associations (Chi-Square)

The Chi-Square tests confirm that every major environmental and operational factor tracked has a statistically significant relationship with survival (all p-values < 0.05).