Naslov
Application of Machine Learning Methods on IoT Parking Sensors’ Data

Autori
Vuk, Dražen ; Andročec, Darko

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of Sixth International Congress on Information and Communication Technology / Yang, XS. ; Sherratt, S. ; Dey, N. ; Joshi, A. - Singapur : Springer, 2021, 157-164

ISBN
978-981-16-2379-0

Skup
6th International Congress on Information and Communication Technology

Mjesto i datum
London, Ujedinjeno Kraljevstvo, 25.02.2021. - 26.02.2021

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Machine learning ; Parking sensor ; Smart city ; Artificial neural networks ; Internet of things

Sažetak
Internet of things brings many innovations and will impact our everyday life. Smart parking is one of the important IoT usage scenarios in urban environments. In this work, we show how to use machine learning methods on Internet of things parking sensors’ data to detect free parking spaces. We have used about 100, 000 instances of data from NBPS parking sensors provided by Mobilisis company. These are actual data from parking sensors with a magnetometer deployed all over the world. The data was preprocessed, normalized, and clustered, because temperature has a large effect on the value of the magnetometer. Next, the XGBoost algorithm and different architectures of artificial neural networks were used to predict whether the parking space is free or not. Used machine learning methods achieve better accuracy than the current classic algorithm based on the history data of a particular parking sensor that is currently used in production (Mobilisis smart parking solution).

Izvorni jezik
Engleski

Znanstvena područja
Informacijske i komunikacijske znanosti

POVEZANOST RADA