Congratulations to Vikky Masih for participating and bagging the Second Prize in the INAI Data Analytics Challenge on Road Safety 2023. His two member team has been awarded a prize money of 50,000 INR. Vikky Masih is a PhD Scholar in our School. We invited Vikky to share with us more about the challenge. Below is the interview transcript.

The challenge was organised jointly by INAI (IIIT Hyderabad), iHub-Data (IIIT Hyderabad), Transportation Research Group of India (TRG) and CSIR-CRRI. The details of the same were notified to all IITG students by Professor Akhilesh Kumar Maurya via email on 27th October 2023. The dataset contained alerts generated by Computer Vision based Collision Avoidance System (ADAS-CAS) and Driver Monitoring System (DMS) installed on various city and inter-city buses running from Nagpur and Hyderabad. The challenge was to do exploratory data analysis and to extract insights and patterns in the environmental factors, road network and driver behaviour. These would be utilised by various stakeholders to improve Road Safety.

The rate of fatal road accidents in India is one of the highest in the world. To reduce the number of accidents, one may introduce a Driver Monitoring System (DMS), and Advanced Driver Assistance System such as Collision Avoidance System (ADAS-CAS). CAS is a pre-crash system that notifies drivers about unsafe situations, while the DMS continuously keeps track and notifies about the driver's alertness. These systems also record the events/notifications generated. Our idea was to take the stream of events thus recorded to extract useful features by first extracting information about each vehicle and trip. Extensive Data Mining was performed on the given dataset while also incorporating secondary datasets such as Road-network information, Weather information, Holidays, Accident Prone Locations, etc.

The most challenging part was to identify hidden patterns in the underlying event generation process as they are influenced by various external factors. For this, the aggregation and usage of secondary data was required. Valuable insights were derived using statistical analysis of the event log after hierarchical clustering of the data. Only a few of the many hypotheses tested resulted in useful insights. The formulation and testing of each hypothesis took considerable effort. Many ideas had to be dropped due to time and human resource constraints.

I would like to thank IIT Guwahati for providing a great environment which facilitates such participation. Exposure to events like these boosts the mental and analytical fortitude of the students. I would also like to appeal to the students to take advantage of similar opportunities which will aid in their professional development.