Objective

The task involves analyzing rides to find solutions that prevent fare overcharging, which often surpasses predicted amounts. By reducing instances of overcharging, we aim to enhance customer satisfaction and loyalty. This effort seeks to address the issue of customers being charged more than expected, improving transparency and fostering stronger relationships with clients, ultimately leading to higher retention rates.

Assumption

• All price variables are expressed in euros.
• All distance variables are expressed in meters
• All duration variables are expressed in seconds
• The change in the price doesn’t include the demand rate on the rider or the payment method
• Distance and duration columns refer to the actual values
• Metered price value means the actual paid amount
• Device name refers to the driver device model

Data Inquiries

In this section, I'll outline the inquiries for which I seek data-driven responses. This involves compiling a comprehensive list of questions to be addressed using available data sources. These questions are crucial for gaining insights and making informed decisions based on the data analysis. They serve as guidelines for structuring the investigation and extracting meaningful conclusions. Through this process, we aim to extract valuable insights that can inform strategic actions and enhance understanding across various domains.

• Does the overcharge happen on weekends, weekdays, holidays, or all of them?
• Does overcharging relate to the number of changed destinations, GPS confidence, and prediction price type?
• Is it caused by a specific app version, rider, or driver device type?
• Is the occurrence of overpaid ride tickets increasing significantly over time?
• Is overcharging linked only to changes in the client's destination? Does the number of changes matter?
• How are price, distance, and duration deviations distributed?
• How do the distributions of rides with significant price deviations compare between those with tickets and those without?
• Validate whether the assumption of a high fraud score for overcharged rides is valid.