Main Article Content
The significant number of fatalities and serious injuries caused by traffic accidents around the world is a worrying problem. Developing nations typically bear a heavier weight of casualties. As a result, developing a model to forecast the likelihood of accidents is extremely difficult. However, the application of machine learning algorithms is one of the significant techniques to forecast the seriousness of such events. As a result, the main goal of the suggested thesis is to automate the process of accident detection by evaluating the severity levels and filtering a set of influential factors that could cause a road accident and generating them using IoT. SMOTE's theoretical notions are put into practice in order to address data imbalance and to ensure that the dataset is balanced. In a later step, the dataset is put to use in the process of building a framework that is constructed from five machine learning algorithms and one stacking algorithm. In the final step of the process, a study is conducted using variables such as the state of the weather and the varying degrees of severity that can have a role in the occurrence of traffic accidents. According to the findings of the experimental analysis that was carried out as part of the research project, the random forest model generated a higher level of accuracy than any of the other models that were put into use, achieving 74%.