Context

Uber relies on accurate driver location data to power dispatch, ETA estimation, and marketplace monitoring. In practice, GPS pings can be incomplete, delayed, or malformed, so analysts need a clear SQL-based approach to handling missing values.

Core question

Explain how you would handle missing values in a driver location dataset stored in PostgreSQL. Your answer should cover:

1. How to identify different kinds of missingness, such as NULL latitude/longitude, empty strings in source fields, or obviously invalid coordinates.
2. How to quantify the issue using SQL so stakeholders understand the scale of missing data.
3. How to decide whether to filter, flag, or impute missing values depending on the downstream use case, such as operational dashboards versus dispatch logic.
4. What SQL patterns you would use to implement the approach safely.

Scope guidance

The interviewer is not looking for advanced modeling or geospatial interpolation. Focus on practical SQL data manipulation: profiling the data, standardizing missing values, creating quality flags, and explaining trade-offs between excluding bad rows and filling defaults.