Fig 1.
Study area and distribution of public transit stations.
Table 1.
Computation of different multi-class models using testing dataset.
Fig 2.
Sample of estimator in selected and trained Random Forest mode.
Fig 3.
Flow chart of supervised modeling for forecasting transit deserts and validation.
Fig 4.
Identified transit gap in peak-time period using aggregated demand.
Fig 5.
Transit gap in peak-time period using aggregation and disaggregation by sex.
Fig 6.
Feature importance in trained Random Forest model.
Table 2.
DICE result.
Fig 7.
(a) Decision plot for samples misclassified as transit desert. (b) Decision plot for samples misclassified as transit oasis.
Fig 8.
(a) Investigation of transit desert measured with male’s transit demand. (b) Investigation of transit desert measured with female’s transit demand.
Fig 9.
(a) Demonstration of the transit desert dashboard displaying an N/A case; (b) Demonstration of the transit desert dashboard displaying a transit desert case.