Predictive Analysis for User Mobility Using Geospatial Data

Abstract

Extremely usage of smart wearable devices such as smartphones and smartwatches which contain various sensors for location detection such as Wi-Fi, LTE, GPS and motion detection such as accelerometer, it has become easier to obtain user mobility data. Today communication systems are becoming more popular due to the developments in communication technologies. There are various services provided which also help to access the data such as video, audio, images from which we can be used to grab the information or pattern of user mobility. The user mobility where user’s movements and locations can be predicted using various methods and algorithms. It can be predicted through data mining, machine learning, and deep learning algorithms where user’s data are fetched from the communication system. A comparative data mining model base on DBSCAN and RNN-LSTM was proposed for predicting the user’s future location-based information predicted from the last locations reported. Mobility prediction based on the transition matrix prediction is done from cell to cell and calculated with the help of the previous inter-cell movement.