Road Accident Analytics using 3 Different Machine Learning Algorithms

• Extracted Crash Incidents using an API call from NHTSA website (US Federal Government Site).
• Crash Data is available from 2010 upto 2022.
• No Personal Identifiers are listed in the entire dataset.
• API is called by iterating the parameters in formatted URL - Year, StateID, Gender, and Injury Severity Level.
• Injury Severity Level is the Dependent Variable(Predictor) and is described using the KABCO scale.
• 9 contributing attributes for an accident and are all Categorical.
• Classification Model is used for building a prediction model.

Code and Resources Used

• Python Version : 3.11.4.
• Packages : pandas, sklearn, matplotlib, seaboard, json, defaultdict.
• API and Data Source: https://crashviewer.nhtsa.dot.gov/CrashAPI.
• Machine Learning Algorithms : Ordinal Logistic Regressor, Naïve Bayes, Random Forest Classifier.