The aim of this study is to provide a data-driven framework for modeling and predicting safe time headway violations on intercity axes using hourly traffic count data from the 141 system. The variable "illegal gap violation," as defined by the system, indicates non-compliance with the minimum two-second time headway from the preceding vehicle and was employed in this study as a safety surrogate measure for proactive risk monitoring. The study domain includes 136 one-directional axes in Razavi Khorasan Province during the year 2025 (1404 Persian calendar). Raw twelve-month data were reprocessed and refined according to a quality control framework encompassing operational duration checks, non-positive traffic volume, invalid speed, class count inconsistencies, estimated count discrepancies, and axis–time key duplication. Out of 1,166,102 hourly records, 1,143,169 records were retained as analysis-ready data. The target variable was defined as the illegal gap violation rate per 1,000 vehicles. Results indicated that the network-wide violation rate was 203.90 violations per 1,000 vehicles. For violation rate prediction, data were aggregated at the axis–month–hour–day-of-week level, with months 1 through 9 used for training and months 10 through 12 for testing. The Random Forest model demonstrated strong performance in temporal testing, achieving an R² of 0.944, an MAE of 13.82, and an RMSE of 21.24. Findings highlight the prominent roles of hourly traffic volume, average speed, overtaking violation rate, and inter-axis differences in explaining the gap violation rate. The results of this research can serve as a foundation for designing early warning systems, ranking high-risk axes, and planning managerial interventions in road safety.