Critical power two wheeler driving patterns at the emergence of an incident

► Bayesian Networks for identifying the critical power two wheeler (PTW) driving patterns. ► High resolution driving data (100 Hz) from a naturalistic PTW driving experiment are used. ► The prevailing actions at the beginning and during critical incidents are identified. ► The methodology efficiently relates the observed actions with different types of incidents. ► The proposed modeling may lead to efficient analysis of the naturalistic data.