Urban transportation constantly seeks a delicate balance: maximizing customer satisfaction and driver utilization while minimizing wait times. This becomes even more challenging during peak demand periods. This project explores how machine learning can help bridge this gap.
I investigated how machine learning can address the following critical research questions:
-
Optimizing Demand and Response: How can machine learning streamline driver dispatch to elevate customer satisfaction and minimize wait times?
-
Dynamic Pricing for a Balanced Ecosystem: Can machine learning-driven pricing strategies balance resource allocation, user affordability, and driver motivation during peak demand periods?
To answer these research questions, I employed two machine learning models:
- Model 1: Ride Demand Forecast Model Our first model is designed to anticipate taxi ride demand data for the upcoming hour, drawing on historical data.
- Model 2: Dynamic Pricing Model The second model is built on the foundation of dynamic pricing, a strategy crucial for real-time price adjustments based on various factors, including demand. This model doesn't operate in isolation; it ingests the predicted demand from Model 1, integrating it with spatio-temporal features that include both ride data and weather data.
