Prediction of learning space occupation through WLAN access point data using Kalman filter and gradient boosting regression

Learning spaces at universities are limited in their capacity, while, providing more such places to students, often imposes quite some problems to the responsible institutions. This leads to problems for the students finding adequate space to conduct their studies on the campus. The consequences are, e.g. large queues of students waiting in front of buildings in the morning, especially during the examination periods, with not many students being able to find a location of their preference. In the course of a day, students change their locations, again searching for new places to sit and learn. In this paper, we present a ML technique that, making use of WLAN access point data in an operational environment of University premises, predicts learning space usage, making use of historical data, and presents them to the student, so they make proper choices. The system has been evaluated on real data and the results are promising for being used in real application environments.