Data Visualization

Short answer: A line chart can show a trend across twelve months at a glance, and a bar chart can rank regions or products. People use visualization for two purposes. Analysts use it to explore data when they look for patterns. People also use it to explain findings, when they prepare a chart for a meeting or a report. The two purposes call for different levels of detail and polish.

Different questions call for different chart types. Line charts suit changes over time. Bar charts suit comparisons between categories. Scatter plots suit the relationship between two measures. Plain tables suit cases where exact values matter. Research on how people read charts has found that people judge positions along a shared scale more accurately than angles or areas. That is one reason bar charts are often easier to read accurately than pie charts. Color, labels, scales and ordering all affect what a reader notices first.

A good chart starts with a question, such as which product lines grew last quarter. The designer chooses the measure, the comparison and the level of detail. Then they pick the chart type that makes that comparison easiest. Bar charts should start their axis at zero, because readers compare bar lengths. Line charts can use a narrower range, because readers compare positions rather than lengths. The chart should state its labels, units and time periods. Use color sparingly to highlight what matters. Do not rely on red and green alone for important differences, since many people find those colors hard to tell apart. Short notes on the chart explain unusual points.

Two warehouses ship 980,000 and 1,020,000 items in a quarter. The difference is 40,000, about 4% of the smaller figure. Suppose a bar chart starts its axis at 960,000. The first bar then shows 20,000 above the start of the axis, and the second shows 60,000. So the second bar looks three times as tall. Starting the axis at zero makes the bars differ by about 4%, which matches the real gap. The manager reading the chart then sees two warehouses performing at a similar level.

Managers usually see data through charts, so chart design shapes which problems they notice and how serious those problems seem. Clear charts speed up meetings and reduce misreading. Anyone approving a dashboard or report can check its axes, time periods and definitions. That confirms each chart shows the data fairly.

Charts can mislead, by accident or on purpose. Common causes are cut-off axes, inconsistent scales, two different axes on one chart, three-dimensional effects and hand-picked time periods. Charts that combine many records can hide important detail, such as one customer driving most of a trend. Charts copied into slide decks often lose their source, date and definitions, so readers cannot check them. Designs that depend heavily on color can shut out readers who have trouble telling colors apart.