AI Hints
Home›Statistics & Math for ML Tutorial›The Normal Distribution
Statistics & Math for ML · Lesson 44 of 138

The 68-95-99.7 Rule

Within 1, 2, 3 SDs.

⏱ 2 min read▶ Video at 50:10🎓 Free
Watch this lesson in the Statistics & Math for ML courseWatch on YouTube ↗Full Statistics & Math for ML course

For any normal distribution, there's a quick rule of thumb. About sixty eight percent of values fall within one standard deviation of the mean. Ninety five percent within two. And ninety nine point seven percent within three.

The 68-95-99.7 Rule
From the video: The 68-95-99.7 Rule
  • 68% within 1 std
  • 95% within 2
  • 99.7% within 3

Checking the rule

We simulate a million adult heights, with a mean of one seventy, and a standard deviation of eight. Then, for one, two, and three standard deviations, we count the share within that distance of the mean. Run it. Sixty eight, ninety five, and ninety nine point seven.

rule.pypython
import numpy as np

rng = np.random.default_rng(0)
h = rng.normal(170, 8, 1_000_000)

for k in [1, 2, 3]:
    inside = np.abs(h - 170) < k * 8
    print(f"within {k} std: {inside.mean():.1%}")
Output
within 1 std: 68.2%
within 2 std: 95.4%
within 3 std: 99.7%

Using the rule

So with a mean of one seventy and a standard deviation of eight, about ninety five percent of people are between one fifty four and one eighty six centimetres. Someone at two hundred and two would be four standard deviations above the mean. That's extremely rare.

keyvalue
±1 std162 to 178 cm, 68%
±2 std154 to 186 cm, 95%
±3 std146 to 194 cm, 99.7%
202 cm4 std, extremely rare

Summary

The sixty eight, ninety five, ninety nine point seven rule tells you how much of a normal distribution is within one, two, and three standard deviations. Next, PDF and CDF.

Key takeaways
  • 68% within 1
  • 95% within 2
  • 99.7% within 3
Next up: PDF and CDF — density and area