Abstract

 Road Traffic Accidents (RTAs) is an alarming cause of many deaths and injuries. It is considered as a major public health issue. Multiple families have been engraved due to RTAs.  Researchers from all over the world are focusing on RTAs because it is a challenging issue. In developed countries, RTAs have been minimized and are being controlled via modern methods of computer science. Similarly, people in Pakistan are facing RTA as their major issue gulping hundreds of lives every day and thousands yearly. In Balochistan province of southern Pakistan, the National Highway 25 (N-25) called Regional Cooperation for Development (RCD) road faces severe RTA issues. According to Medical Emergency Response Center (MERC) overall 25,033 accidents have been recorded from Oct-2019 to Oct-2022 with 33,726 injuries and 653 deaths. This shows the alarming condition of N-25 for public and also serious challenges for the traffic controlling authorities. In Khuzdar area, poses every year more than five thousand RTAs happen taking around 250-300 lives and two hundred injuries having 500 to 600 accidents disabled person (ADPs). However, other studies have been focused on generally on fuzzy methods but this study has been used five defuzzification methods which is the new contribution in to the field of fuzzy logics and Artificial intelligence tools. In this research, we have presented a Fuzzy Inferencing System (FIS) for investigating the RTAs with a case study specific to Khuzdar area from Baghbana to Pir Umar. A comparison of five defuzzification methods, namely centroid, bisector, Smallest of Maximum (SOM), Large of Maximum (LOM) and Medium of Maximum (MOM) was made. The results indicated that Large of Maximum (LOM) method is able to provide scalable output which is better in reflecting the ambiguities of the related variables, their membership functions, and associated rules. After comparison of defuzzification method it was seen that these methods can be useful in reduction of the RTAs which will also help the traffic authorities of N-24 MERC, police and other controlling authorities. This study can be useful for the development of other AI methods and a way forward for other researchers.