Percentiles and Quantiles from CDFs
What you'll learn: How to extract percentiles from continuous distributions by "reversing" the cumulative distribution function.
The Core Idea
You already know that a CDF tells you the probability of observing a value less than or equal to some number. A percentile or quantile flips this question around: instead of asking "what's the probability below x?", you ask "what value of x has a given probability below it?"
Mathematically, if F(x) = P(X ≤ x) is your CDF, then the p-th quantile (or 100p-th percentile) is the value x where F(x) = p.
Finding Percentiles: Inverting the CDF
Think of the CDF as a machine that converts values into probabilities. The quantile function (also called the inverse CDF or percent-point function) reverses this: it converts probabilities back into values.
For example:
- The 50th percentile (median) is the value where
F(x) = 0.5 - The 95th percentile is where
F(x) = 0.95 - The 25th percentile (first quartile) is where
F(x) = 0.25
Practical Interpretation
If you know the 90th percentile of exam scores is 85, it means 90% of students scored 85 or below (equivalently, only 10% scored above 85). You found this by solving: "What score has a CDF value of 0.90?"
This is incredibly useful for setting thresholds, understanding risk (e.g., "worst 5% of outcomes"), and comparing positions across different distributions.
Key Takeaway: Percentiles invert the CDF—given a probability p, the p-th quantile is the x-value where the cumulative probability equals p, effectively answering "what value cuts off the bottom p% of the distribution?"