Accident Severity Prediction in Big Data Using Auto-Machine Learning

Estimating the severity of a traffic accident is problem in motor vehicle because it affects saving human life. If value can be predicted before occurs, all emergency teams needed could sent to area provide faster first aid. With this aim, we studied big data set for accidents USA between 2016 and 2020, which almost 2.25x106 rows long. First, preprocessed by removing unnecessary variables. Then with blank cells are removed. Finally, about 1.7x106 length left prediction process. A machine learning algorithm has been used determine classification based on 16 input parameters. Moreover, binary decimal count conversation as novel preprocessing method. As result, model built total accuracy 0.816. test results also validated precision, recall, f1-score values. In study, an auto-machine developed trained predict possible weather road conditions.