INDUSTRY:
SHARED MICROMOBILITY
YEAR:
2024
EXPERIENCE:
PREDICTIVE MODELING
CAPITAL BIKE-SHARE
about.
Capital Bikeshare Demand Modeling
This project explored the key factors influencing daily demand for the Capital Bikeshare system using ridership and weather data from 2023. I aggregated over 4 million raw trip records into structured daily counts segmented by rider type, bike type, and location. Using regression and machine learning models, I analyzed how user attributes and weather conditions affected ridership volume, with a focus on interpretability and predictive performance.
result.
The analysis revealed that classic bikes, round-trip rides within D.C., and long-term members were most strongly associated with higher daily ridership. Temperature, UV index, and precipitation also had measurable effects on demand. The final Random Forest model outperformed other approaches, achieving a cross-validated RMSE of 270 and explaining over 94% of the variance. These insights support operational planning and rider engagement strategies for bikeshare systems seeking to scale efficiently in urban environments.
