Idea
An explainable AI model predicting passenger wait times before ride requests to improve ridesharing user experience and operations.
Research Paper
Core Innovation
This paper introduces FiXGBoost, a novel feature interaction-based XGBoost model that predicts passenger waiting times before ride requests without driver matching data. It uniquely quantifies the importance of demand and supply factors, enabling explainable and accurate predictions. This approach advances prior work by focusing on pre-request prediction and interpretability at scale.
Market Size (TAM)
$10–20B TAM, $2–5B SAM; assumption: ridesharing and urban mobility markets growing with demand for predictive analytics.
Potential Customers & Pain Points
- Ridesharing Platforms Needing Accurate Wait Time Estimates
- Urban Mobility Planners Seeking Demand-Supply Insights
- Passengers Wanting Reliable Wait Time Predictions
Business Model
Licensing predictive API to ridesharing platforms and urban planners; offering analytics dashboards for operational insights.
Competitive Landscape
- Uber ETA Prediction
- Lyft Wait Time Models
- Google Maps Transit Predictions
Implementation Challenges
- Data Privacy and Access Limitations
- Integration with Existing Ridesharing Systems
- Model Generalization Across Cities
Validation Strategy
- Pilot integration with a mid-size ridesharing platform
- Compare prediction accuracy against existing benchmarks
- Collect user feedback on wait time reliability improvements
Research Paper Overview
A First Look at Predictability and Explainability of Pre-request Passenger Waiting Time in Ridesharing Systems
Summary
This paper studies pre-request passenger waiting time prediction in ridesharing, focusing on predicting wait times before ride requests without knowing matched driver info. It analyzes demand and supply dynamics, proposes FiXGBoost, a feature interaction-based XGBoost model, and quantifies factor importance. Experiments on 30M+ trip records show FiXGBoost achieves accurate, explainable pre-request waiting time predictions.