- Project
- 2023
- Tools: Python
- full project
The project focuses on achieving the following objectives:
- Analyze and visualize patterns in motor vehicle collisions.
- Develop machine learning models to predict collision risk levels.
- Provide recommendations based on the analysis to improve road safety.
The dataset, sourced from NYC Open Data provided by the Police Department (NYPD), contains detailed information on motor vehicle collisions. Key variables include geographical information, contributing factors, vehicle types, and injury/fatality counts.
The dataset undergoes rigorous preprocessing, including standardizing column names, converting time-related features, creating new features like total injuries and killed, risk level categorization, and handling missing values through removal and imputation.
EDA explores the distribution of contributing factors, vehicle types, and the relationship between accidents and time variables such as months and hours. Visualizations include count plots, log-scale distributions of injuries and fatalities, and line plots showing the frequency of accidents over time by borough.
The project employs three distinct modeling approaches:
- Cluster Analysis Model: Utilizing K-means clustering to identify major vehicle types associated with collisions.
- Linear Regression for Accidents Prediction: Predicting the number of accidents in an area over time using linear regression.
- Decision Tree Classifier: Classifying the risk level associated with accidents based on factors such as vehicle types, contributing factors, and counts of injuries and fatalities.
- The cluster analysis identifies three major clusters, each associated with specific vehicle types involved in collisions.
- The linear regression model predicts the number of accidents over time, providing insights for proactive planning.
- The decision tree classifier categorizes accidents into risk levels, aiding in understanding factors contributing to severity.
The project concludes with actionable recommendations, including identifying high-risk areas, improving road infrastructure, encouraging sustainable transportation, enhancing emergency response, promoting vehicle safety technologies, implementing safety policy measures, and enforcing stricter vehicle standards inspection.
While acknowledging room for refinement, the project contributes valuable insights to understanding collision patterns, emphasizing the potential impact on road safety, and providing a foundation for further improvements in predictive models and safety measures.