Why taxi orders fail? EDA 🚗

Download dataset and source code here

Expected Outcome

• Understand Failure Reasons: Identify which failure reasons are most common.

• Identify Critical Times: Spot times with high failure rates and their causes.

• Optimize Cancellation Processes: Improve processes based on average cancellation times.

• Improve ETA Predictions: Refine ETA predictions to reduce failures.

• Geographical Insights: Visualize problem areas to target improvements.

By achieving these goals, we aim to enhance the overall efficiency of the Gett platform, leading to better service for customers and more successful order completions.

Insights:

1. Distribution Analysis of Failures

   

   Which reason for failure is most common --> Cancelled by Client

   Number of orders that failed for each reason?

   • • Status Key 4 (Cancelled by Client): 7,307 orders

       • Status Key 9 (Rejected by System): 3,409 orders

2. Failure Trends by Hour

   

   • Specific times of the day when orders fail more frequently -> 8am

3. Average Cancellation Time Analysis

   

   • Task: Compare the average time it takes for cancellations with and without a driver assigned.

   • Deliverable: A plot of average cancellation times by hour.

   • Outcome: Detect any unusual cancellation patterns and outliers.

4. ETA Distribution Analysis

   

   • Task: Examine the average estimated time of arrival (ETA) for failed orders.

   • Deliverable: A plot showing how ETA varies by hour.

   • Outcome: Understand how ETA influences order failures.