Analytics
Cancellation Policy for Ridesharing
A/B testing and root-cause analysis on large-scale ridesharing data
Overview
An analytics project that designs and evaluates a cancellation policy using A/B testing, root-cause analysis, and statistical validation.
Problem
Cancellation behavior can reduce platform reliability and revenue, but policy changes need causal evidence before rollout.
Dataset
A ridesharing dataset with more than 1M rows covering cancellation behavior, trip context, and treatment groups.
Approach
Used Pandas and Matplotlib for root-cause analysis, designed three treatment groups, and validated policy impact with SciPy and Statsmodels.
Results
Estimated roughly $6M in annual revenue savings under the proposed policy and identified key cancellation drivers.
Lessons Learned
Operational analytics needs both statistical significance and business framing so that recommendations connect to platform decisions.
Model / Pipeline
The implementation combines Python, Pandas, A/B Testing, SciPy, Statsmodels, Matplotlib with a repeatable workflow for data preparation, evaluation, and communication.