Fig 1.
Exemplary itinerary intersecting the hexagonal grid in Greater London (anonymised) © OpenStreetMap contributors.
Fig 2.
Correlation plots of trips variables in Greater London (A) and Brisbane (B).
Fig 3.
Variable importance plots for Greater London (A) and Brisbane (B).
Distance is the most important variable in both cities and noise is more relevant to the mode of travel prediction in Greater London than in Brisbane.
Fig 4.
Partial plots for Greater London (A) and Brisbane (B).
The log odds of predicting a trip to be made by different mode of transportations (car, bike or walking) broke down by each contextual variable (distance, street lengths, education, recreation, health, retail, building area) and trip share travelled across each noise class.