Your baby is in the 75th percentile for height. Your test score is the 90th percentile. Neither is a percentage *of* anything — it's a rank: the fraction of everyone you're above.
The kth percentile is the value below which k% of the data falls. The 50th percentile is the median; the 25th and 75th are Q1 and Q3.
Growth charts, standardised test scores, income brackets, response-time SLAs ('99th percentile latency'), grading curves — percentiles describe *position*, not magnitude.
A high percentile means you're above most of the data. The 100th percentile is the maximum; the 0th is the minimum.
A test has 500 students. You scored higher than 425 of them. What percentile?
What does '95th percentile latency = 200ms' mean for a website?
95% of requests completed in 200ms or less; the slowest 5% took longer. Engineers track *high* percentiles because the average hides the painful tail.
A percentile is a rank, not a score or a percentage. Being in the '90th percentile' on a test doesn't mean you got 90% — it means you beat 90% of test-takers.
From a cumulative-frequency curve (ogive), read percentiles directly: go up to k% of the total count, across to the curve, down to the x-axis.
- kth percentile = value below which k% of the data falls.
- Median = 50th; Q1 = 25th; Q3 = 75th percentile.
- It's a rank/position, not a score.