Advanced computing to support urban climate neutrality

Table 2 MAE computed between the measured and predicted number of vehicles using various ML models

Model	Year	MAE	Std MAE
Baseline (Mean)	16–17	39.412	0.000
Baseline (Mean)	17–18	40.102	0.000
Baseline (Mean)	18–19	39.945	0.000
Baseline (Mean)	19–20	40.891	0.000
Linear regression	16–17	18.042	2.165
Linear regression	17–18	19.532	2.277
Linear regression	18–19	17.924	2.383
Linear regression	19–20	17.642	2.014
Random forest	16–17	12.624	2.982
Random forest	17–18	22.632	6.232
Random forest	18–19	16.832	2.945
Random forest	19–20	13.989	2.593
Neural network	16–17	9.192	1.082
Neural network	17–18	11.892	1.347
Neural network	18–19	10.720	1.998
Neural network	19–20	10.204	0.934
AutoGluon	16–17	8.623	1.065
AutoGluon	17–18	10.792	1.492
AutoGluon	18–19	10.261	1.890
AutoGluon	19–20	9.409	0.952

The evaluation was conducted on data from a previously unseen year. Each model underwent 20 rounds of training and evaluation across four distinct train/test splits

ISSN: 2192-0567