Development of crash prediction models by assessing the role of perpetrators and victims: a comparison of ANN & logistic model using historical crash data.

Road traffic injuries cost countries 3% of their annual GDP. In developing countries like India, every year around 150,000 people die on roads. The type of vehicles involved in a crash contribute majorly to the outcome of casualty (injury/death). Barring few studies, literature are less regarding the role of vehicle as perpetrator and victim on road crash fatalities. Historical crash data has been used in the present study to examine the role of vehicles (both as perpetrator & victim). The study reveals that victim's effect is more as compared to perpetrator/accused for determining the outcome of crash. Heavy vehicles as perpetrator, and self-hitting vehicles along with pedestrians as victims have higher fatality rates. Binary logistic regression and artificial neural network (ANN) has been utilized for developing prediction models. Binary logistic model predicted around 75% of outcomes correctly with default cut-off value (0.5). However, based on reported crash data, where 19% of total crashes lead to deaths, 0.19 has been proposed as cut-off value which increases the accuracy of the predictions. Accuracy of ANN technique directly depends on the number of crashes reported for a definite pair of perpetrator and victim and the type of validation technique used (Holdback/K-Fold) along with the type of hidden layer chosen for the study based on different types of sigmoid activation function. ROC curves in ANN suggest that the analysis can predict 75% of the outcomes which can be increased by deleting the pairs of vehicles which are present/have occurred in very less number. A comparison has been made between the two techniques based on their advantages and limitations. The developed models can be used as safety indicators based on composition of traffic flow on urban roads.

Factors propelling fatalities during road crashes: A detailed investigation and modelling of historical crash data with field studies.

One of the major concerns in developing countries like India is to maintain traffic safety under mixed and heterogenous scenario. Although zero accidents is the need of the hour, the first step to attain it is ensuring zero deaths and no serious long-term disabling injuries in road crashes. To reduce the road crash fatalities, explicit and detailed studies have been conducted by utilising historical road crash data of two emerging smart cities of India - Bhubaneswar and Visakhapatnam. Traffic flow data and characteristics of road infrastructure has also been collected by performing field studies at accident prone locations. Various factors including vehicular characteristics, road user characteristics, and road infrastructure have been analyzed using various non-parametric tests to identify the contributing factors resulting in fatalities. It is observed that out of 14 variables used for study, 8 factors were significantly related to fatal crashes. These included categories of victim and accused, 85th percentile speed, presence of road markings, availability of sight distance, etc. The significant factors were subjected to binary logistic regression to determine the odd's ratio of significant factors. The logistic regression predicted 79% of deaths correctly. Crash fatality prediction models are developed using both Classification and Regression Tree (CART) classification tree with 83% accuracy. Although CART classification led to higher accuracy, binary logistic regression is more robust as it considered more significant factors as compared to CART. Subsequently, a severity index has been proposed based on proportions of actual fatal crashes and usage of K-means clustering technique. The proposed indices shall be really helpful in traffic safety management, specifically in reduction of fatalities during road crashes.