Identification of Mobile Traffic Characteristics in University Area with Decision Tree Learning Model

In order to accommodate the explosively increasing mobile data traffic, many attempts have been made to increase mobile communication resource allocation efficiency. One of them is to create a system that predicts data traffic by using machine learning or deep learning models. If mobile communication resources are allocated in advance by predicting the time when traffic surges, data transmission can be performed smoothly. Therefore, more accurate traffic prediction is required, and traffic prediction accuracy is improved by clustering regions in various ways. In this paper, I checked the characteristics of traffic for each specific place using the university as an example. And places with similar traffic characteristics found commonalities using statistical data. If the characteristics of each specific place are confirmed, the accuracy of traffic prediction can be dramatically improved by grouping places with the same characteristics and clustering them. In order to extract only the characteristics of the place, the influence of regional and cultural characteristics on traffic had to be deleted. So, we took university traffic datasets from India and Milan, Italy, and analyzed them by finding commonalities in regions with different cultures. And it was confirmed that the supervised learning model called Decision Tree can identify the characteristics of university traffic grid by itself and classify regions that show the same characteristics.