An anlysis of different models that can be used to score routes based off of safety/past crash data for the navigAid application: https://github.com/Rayna-Yu/NavigAid
This repository contains:
- Data Preprocessing Scripts: Methods for cleaning and transforming raw crash data into usable formats.
- Original Datasets: Access to the unprocessed crash data used in our analysis.
- Cleaned and Processed Datasets: Refined datasets ready for model training and evaluation.
- Model Implementations: Code for various machine learning models applied to predict route safety.
- Evaluation Metrics: Scripts to assess model performance and compare different approaches.
We explored a couple of different models for classification including:
- Logistic Regression
- SVM (support vector machine)
- Gradient Boosting
- Random Forest
Ultimately, we dove the most in-depth with evaluation metrics for Random Forest for accuracy and interpreability. Feel free to look around and try it out!