“…The methods found in the literature can be divided into a few overlapping groups of approaches. Compared to the related work section, we excluded a group of methods incorporating short-term dynamics and included a time series forecasting approach that estimates the number of people over the whole map at a specific time: - 1) spatial-only models ( Section 2.1 ) that do not take time into account ( Kucner et al, 2016 ; Senanayake and Ramos, 2018 ),
- 2) time series forecasting methods that do not take the structure of the space into account ( Vintr et al, 2018 ),
- 3) partially discrete and partially continuous models, ( Section 2.3.3 ), which incorporate continuous models in the cells of a predefined grid ( Krajnik et al, 2014a ; Molina et al, 2018 ).
- 4) methods that model spatial and temporal features separately ( Section 2.3.1 ) and understand them as independent ( Bennetts et al, 2019 ; Kubiš, 2020 ),
- 5) continuous spatio-temporal methods ( Section 2.4.2 ) that model the spatio-temporal phenomena together ( Vintr et al, 2019b ; Krajník et al, 2019 ), and
- 6) continuous spatio-temporal methods ( Section 2.4.1 ) that model the temporal evolution of the continuous spatial model ( Zhou et al, 2015 ; Zhou and Matteson, 2016 ).
…”