الفهرس 
الغلافOnly 14 pages are availabe for public view
Abstract
It is very important for traffic management to be able to correctly recognize traffic trends from large historical traffic data, particularly the congestion pattern and road collisions. This can be used to reduce congestion, improve protection, and increase the accuracy of traffic forecasting. Text Mining is one of the most critical techniques of processing and analyzing unstructured data especially since more than 80% of today’s data is composed of unstructured or semi-structured data, and it is almost impossible to manage. Text mining uses Natural Language Processing (NLP) techniques to extract valuable insights and useful information from large volumes of unstructured text data. Choosing the correct and effective text mining methodology helps speed up and reduces the time and effort needed to retrieve valuable knowledge and information for future prediction and decision-making processes. Using different classification and machine learning techniques multi algorithms applied to get the optimum classifiers used in our model. Such as (Random Forest, Decision Tree, Support Victor Machine (SVM), K-Nearest Neighbor (K-NN), Logistic Regression, and Naïve Bayes). The experimental results on real-world datasets demonstrate that our models achieved results better than Prayag Tiwari’s Research Work related to the Leeds UK Dataset. on the other side in our thesis, more classifiers applied to classify this dataset based on casualty class and severity class, so we can see clearly that what circumstances affect and who is involved more in an accident between the driver, passenger, or pedestrian. In addition to Prayag Tiwari’s Research Work elements, the severity class added which is considered an added contribution selected for high traffic management perspective. The other contribution which considers an added value in our model is utilizing more classifiers algorithms which are (Random Forest and Logistic Regression). Already The better classifier in our work regards Casualty Class is Random Forest classifier which achieved accuracy 87.79% better than others and lowest classifier accuracy is Logistic Regression which achieved accuracy 75% and the Decision Tree is better classifier in our thesis regards Casualty Severity which achieved accuracy 88.02%, and lowest classifier accuracy is Naïve Bayes which achieved accuracy 86.35%..