Data Anomaly Hunting

A framework for finding meaningful insights in data by looking for unexpected patterns or anomalies that spark curiosity, as shared through an example from Uber.

Core Concept

• Don't always analyze data with a specific answer in mind
• Look for things that make you go "that's weird" or seem counterintuitive
• Search for data points that don't follow expected trends

Example: Uber Ride Analysis

• Key anomaly discovered:
  • 60%+ of rides were single passengers
  • 60%+ of trips were less than 3 miles
  • Most rides used 4-6 passenger vehicles
• The "weird" insight:
  • Using large multi-passenger vehicles for predominantly single-passenger, short trips seemed inefficient
  • Led to business opportunity: developing smaller, single-passenger vehicles optimized for short urban trips

Application Method

• Browse data openly without specific queries in mind
• Look for patterns that seem misaligned or surprising
• When something stands out, dig deeper to understand why
• Use these insights to identify potential business opportunities or areas for optimization

Key Takeaway

• Sometimes the most valuable insights come from finding what doesn't make sense, rather than confirming what you expect to see
• Data anomalies can reveal untapped business opportunities or inefficiencies in current systems