Predicting Bike Demand in Washington, D.C.
A time-series model forecasting Capital Bikeshare demand in 15-minute intervals
A major challenge for bikeshare systems is optimally redistributing bikes across the network — a station without bikes or without open docks inconveniences riders and pushes them toward other transportation. This project develops a predictive model and time-series analysis for Capital Bikeshare in Washington, D.C., forecasting demand in 15-minute intervals for the upcoming week. Rather than relying on resource-intensive truck-based resupply, the model supports a dynamic-incentive approach — encouraging riders themselves to redistribute bikes across the network.