🚑 Road Accident Survival Analysis (Python EDA)

---

Overview

This project explores factors associated with survival in road accidents using a structured tabular dataset.
The analysis is implemented end-to-end in a Jupyter Notebook using Python, Pandas, NumPy, Matplotlib, and Seaborn.

The main goal is to understand how age, gender, speed, helmet use, and seatbelt use relate to the likelihood of surviving a road accident, and to practice a clear exploratory data analysis (EDA) workflow suitable for real-world analytics roles.

---

Business Motivation

Road accidents are a major public health problem. Survival often depends not only on the severity of the crash but also on human behaviour and safety equipment usage.

By analysing an accident survival dataset, this project aims to:

• Quantify how survival varies across demographic groups.
• Compare survival outcomes for people who used helmets or seatbelts versus those who did not.
• Build evidence that can support policy, education, and enforcement decisions in road safety.

---

Dataset

The dataset contains one row per person involved in a road accident.
Key columns used in the analysis include:

• Age – Age of the individual.
• Gender – Gender of the individual.
• Speed – Reported speed at the time of the accident.
• Survived – Survival outcome (e.g. 1 = survived, 0 = did not survive).
• Helmet_Used – Whether a helmet was used.
• Seatbelt_Used – Whether a seatbelt was used.

Note: the notebook focuses on cleaning, exploring, and visualising these core features rather than building predictive models.

---

Methodology

1. Data Loading & Inspection

• Loaded the CSV file into a Pandas DataFrame.
• Inspected structure using .head(), .info(), and .describe().
• Verified column types and basic distributions.

2. Data Cleaning

• Removed records with missing values (dropna) to ensure consistent analysis.
• Checked data types for each column (numeric vs categorical).
• Performed simple sanity checks (e.g. reasonable age and speed values).

3. Exploratory Data Analysis (EDA)

Core questions explored:

• Age distribution of accident victims.
• Number of accidents by gender.
• Overall survival distribution (how many survived vs did not).
• Minimum and maximum speeds recorded.
• Counts of accidents involving helmet use.
• Counts of accidents involving seatbelt use.
• Survival rates by gender.
• Survival rates by helmet and seatbelt usage.

Visuals created in the notebook include:

• Histogram of Age
• Bar charts of Gender, Helmet_Used, and Seatbelt_Used
• Pie charts for overall survival distribution
• Bar charts for survival counts by category

These plots give a fast, interpretable view of how different factors are related to survival.

---

Key Analytical Highlights

• A full exploratory workflow is implemented: loading → cleaning → summarising → visualising.
• Survival is analysed from both a demographic perspective (age, gender) and a behavioural/safety perspective (helmet and seatbelt use).
• The notebook computes survival rates grouped by key variables, enabling comparisons such as:
  • survival percentage by gender
  • survival percentage for helmet vs no-helmet
  • survival percentage for seatbelt vs no-seatbelt
• Visualisations make the findings easy to communicate to non-technical stakeholders.

This project is intentionally focused on exploratory analysis and insight generation, which is a critical step before building any predictive model.

---

Recommendations & Next Steps

Recommendations

• Use survival-by-equipment analysis to support stronger helmet and seatbelt adoption campaigns.
• Target road safety education at demographic groups that appear more frequently in severe or non-survival cases.
• Combine this dataset with additional contextual data (e.g. time of day, road type, vehicle type) to improve understanding of high-risk scenarios.

Potential Extensions

• Build a logistic regression model to predict survival probability based on available features.
• Introduce statistical tests (e.g. chi-square tests for categorical associations).
• Engineer new features such as speed bands (low/medium/high) or age groups.
• Deploy a simple Streamlit or Dash dashboard to allow interactive exploration by policy makers or analysts.

---

Tools & Technologies

• Language: Python
• Environment: Jupyter Notebook
• Libraries:
  • pandas, numpy
  • matplotlib, seaborn
  • PIL / IPython.display for loading cover image

---