The shape of data: distributions and histograms

What you’ll learn

This is lesson 3 of Track 9 (Statistics & Probability for AI) and the third lesson of Phase 1 (Describing data). The previous lesson squeezed a dataset into a center and a spread; this one shows what those numbers throw away. Two datasets can share a mean and a standard deviation and still look nothing alike, because summary numbers cannot capture shape. You will build the most useful picture in statistics, the histogram, learn to read the handful of shapes that cover most real data, and see why looking at a distribution catches problems that numbers hide. The source curriculum is Khan Academy’s Statistics & Probability course, by Sal Khan and the Khan Academy team, freely available and cited as further study.

The lesson builds a histogram from raw numbers, shows how bin width changes the story, names the shapes you will meet (symmetric, right- and left-skewed, uniform, bimodal, and the bell curve), reconnects skew to the mean-versus-median gap from the previous lesson, and closes on why a practitioner inspects the distribution of every feature and label before trusting it.

Where this fits

This is lesson 3 of 14, the third lesson of Phase 1. The previous lesson, Summarizing data: center and spread, gave the numbers; this lesson gives the picture and shows how skew (a disagreement between mean and median) appears as a tail. The bell shape introduced here gets its own full treatment later in the track, in the normal-distribution lesson, and class imbalance (visible in a histogram of labels) ties straight back to the base-rate idea from lesson 1.

Before you start

Prerequisites: the previous lesson (Summarizing data: center and spread), since this lesson connects shape back to the mean and median. No new arithmetic is required; the work here is reading pictures, not computing.

About the math

There is essentially no computation in this lesson. It is about reading shapes: building a histogram by counting into bins, then recognizing symmetric, skewed, uniform, bimodal, and bell shapes by eye. The histograms are drawn in plain text so you can read them anywhere, and the one quantitative idea (how the mean and median shift with skew) is the carry-over from the previous lesson, shown as a picture rather than a calculation.

By the end, you’ll be able to

Explain what a histogram is and how bin choice changes the picture
Name and recognize the common distribution shapes (symmetric, skewed, uniform, bimodal, bell-shaped)
Connect a right- or left-skewed shape to the gap between the mean and the median
Explain why a histogram reveals problems (outliers, two populations, class imbalance) that summary numbers hide
Describe why inspecting feature distributions is a routine first step before machine-learning modeling

Time and difficulty

Read time: about 11 minutes
Practice time: about 13 minutes (a self-check, a name-the-shape-and-predict-mean-vs-median exercise, a spot-the-problem exercise on feature distributions, and flashcards)
Difficulty: standard (visual and conceptual; no new computation)