An outlier is an observation that differs markedly from most others. It can pull the mean away from what feels typical, while the median is less affected. section 2.14
Useful clues for spotting outliers include:
- A large gap between mean and median suggests that extreme values are skewing the data. section 2.14
- Under the empirical rule, observations beyond two or three standard deviations from the mean are often treated as unusual. section 4.7
- Extreme observations deserve scrutiny rather than automatic dismissal: they may reveal unusual cases or problems with a model. section 4.7
The high-roller bar tab
For bar tabs of 7, 18, 19, 20, 21, 22, 25, 120, and 1,500 dollars, NumPy reports a mean of about $194.67 and a median of $21. The $1,500 tab is an outlier, so $21 better represents a typical patron. section 2.14
