Seeing the Data
A well-chosen graph reveals patterns that a table of numbers hides. Data visualisation matches the type of data to the right display, so structure — shape, trend, relationship, outliers — becomes visible at a glance.
The core chart types and their purposes:
- Histogram — the distribution of a single numerical variable, grouping values into bins whose bar heights show frequency. It reveals shape (symmetric, skewed), spread, and outliers.
- Bar chart — compares a numerical value across categories (distinct, unordered groups). Bars have gaps, unlike a histogram's touching bins.
- Scatter plot — the relationship between two numerical variables, each point a data pair. It exposes correlation, clusters, and outliers in two dimensions.
- Line chart — a value over an ordered sequence, usually time (a trend).
- Box plot — a compact summary of a distribution via its five-number summary (min, quartiles, max), excellent for comparing groups and spotting outliers.
The crucial distinction is histogram versus bar chart, which look similar but display different data. A histogram bins a continuous numerical variable — the horizontal axis is a number line, and bars touch because the bins are adjacent intervals. A bar chart compares categorical groups — the horizontal axis has distinct labels, and bars have gaps because the categories are separate. Using one where the other belongs misrepresents the data.
Good visualisation also means honest design: a truthful axis (usually starting at zero for bar heights), no distorting 3D effects, and a scale that neither exaggerates nor flattens the real differences. A misleading chart can distort the same data as badly as a wrong number.
Common pitfall: confusing a histogram with a bar chart. A histogram shows the distribution of a numerical variable (binned, bars touching, axis is a number line); a bar chart compares categories (distinct labels, bars separated by gaps). They look alike but answer different questions — using a bar chart for continuous data, or a histogram for categories, misrepresents what the data actually shows.