Accident Data of Selected Indian Highways for Accident Severity Prediction Using Machine Learning Models

The dataset utilized for this study comprises road accident data collected from selected stretches of highways under the jurisdiction of the National Highways Authority of India (NHAI). The data was provided by the respective Concessionaires and covers four major highway projects across multiple states, representing diverse geographic, traffic, and environmental conditions. This variety of data is crucial for developing a comprehensive and generalizable model for predicting road accident severity. Four distinct highway segments under NHAI jurisdiction were chosen to ensure geographic and operational diversity:1. Pune–Solapur (NH‑9, km 144 400–249 000, Maharashtra): 2,804 records2. Barwa‑Adda–Panagarh (NH‑2, km 398 240–521 120, Jharkhand & West Bengal): 3,710 records3. Chengapally–Walayar (Chainages 102 035–144 680 & 170 880–183 010, Tamil Nadu): 422 records4. Nagpur Region (Maharashtra): 1,180 records