Last Thursday, February 11, Jamie Arjona, industrial doctoral student of inLab FIB, presented his thesis “Design smart its services through innovative date analsys modeling”, directed by Josep Casanovas and Ma Paz Linares, which obtained a final result of excellent.
Jamie’s thesis is the product of a collaboration between inLab FIB and the company Worldsensing S.L. with the aim of characterizing and modeling different parking systems from around the world using advanced modeling techniques. Worldsensing S.L. are manufacturers of sensors for industrial and Smart Cities and also offer different software solutions for these areas. The sensors used provide data on the occupancy status of parking spaces in real time. This data is stored and subsequently analyzed to create the models.
The models chosen are based on the neural network methodology, specifically the Multilayer Perceptron (MLP), Long Short Term Memory (LSTM) and Gated Recurrent Units (GRU) models, and a comparison has also been made with classical models such as the Autoregressive Integrated Moving Average (ARIMA).
In addition, a study has been carried out using other data sources and analyzing their effect on the parking processes and quantifying how this information improves the predictions of the models. These data are climatic effects and calendar events.