baqapp - Accident Prediction in Barranquilla

Project Overview

This project is an investigation into traffic accidents in Barranquilla, Colombia, with the aim of developing predictive models to identify accident hotspots and contributing factors. By analyzing a dataset of 25,610 accident records, this research seeks to provide insights that can guide urban planning and road safety interventions.

Project Structure

The project is organized as a Quarto website, with the following structure:

accident-prediction/
├── introduction.qmd
├── data-exploration.qmd
├── methodology.qmd
├── results.qmd
├── conclusion.qmd
└── _quarto.yml

Files and Directories

• introduction.qmd: Provides an introduction to the study, including background, justification, and model selection.
• data-exploration.qmd: Contains an initial exploration of the dataset, including summary statistics and visualizations.
• methodology.qmd: Describes the methodology used, including data cleaning, feature engineering, and model selection.
• results.qmd: Presents the results of the predictive modeling, including performance metrics and key findings.
• conclusion.qmd: Concludes the study with a summary of findings and implications for road safety.
• _quarto.yml: Configuration file for the Quarto website.

Dataset

The dataset used in this study contains 25,610 records of traffic accidents in Barranquilla, with the following columns:

• FECHA_ACCIDENTE: Date of the accident
• HORA_ACCIDENTE: Time of the accident
• GRAVEDAD_ACCIDENTE: Severity of the accident
• CLASE_ACCIDENTE: Type of accident
• SITIO_EXACTO_ACCIDENTE: Exact location of the accident
• CANT_HERIDOS_EN_SITIO_ACCIDENTE: Number of injured at the accident site
• CANT_MUERTOS_EN_SITIO_ACCIDENTE: Number of fatalities at the accident site
• CANTIDAD_ACCIDENTES: Number of accidents
• AÑO_ACCIDENTE: Year of the accident
• MES_ACCIDENTE: Month of the accident
• DIA_ACCIDENTE: Day of the week of the accident

Model Selection

The primary model used in this study is Random Forests, chosen for its balance between predictive accuracy and interpretability. This model is well-suited for handling complex datasets and has been effectively used in similar traffic accident prediction studies.

How to Build and View the Project

To build and view the Quarto website locally:

1. Clone the repository:

   git clone https://github.com/neisser/baqapp.git

2. Navigate to the project directory:

   cd baqapp

3. Install Quarto (if not already installed). Instructions can be found here.

4. Preview the website locally:

   quarto preview

5. The website will be available at http://localhost:4321.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

Special thanks to the contributors of the dataset and the academic community whose research informed this project.