Bar chart

From Business Intelligence

Tufte-inspired bar chart. No vertical axis, no grid lines, no tick marks, and a single color for all the bars. It also uses the pareto principle, which arranges the bars from the smallest to largest values.

This Tufte-style bar chart strips out the chartjunk, and moves the tick marks inside the bars. However, bar charts should not be used to depict time series. From Excel charts designed by William Oswald

The training courses offered through DOH are useful for my current position graph is well-designed, but the data density is very low. A dataset this small is better presented in a table.

Description

Classic method for comparing a set of values using two data series and a label. It features either horizontal or vertical bars that represent the quantitative value of an element. Good for showing quantitative differences and making quick comparisons between several data items of the same class.

Data requirements

Two data series (x and y) representing discrete data.

Availability

SSRS, Business Objects

Considerations

Bar charts represent data as discrete units and should not be used to represent change or a time series. A line graph is better for this.
Bar charts with a long list of sideways labels on the x-axis should consider rotating the chart vertically, using a simple table instead, or finding ways to abbreviate.
Often the grid lines and ticks are not needed on a bar chart. Try using very light gray lines or eliminate them altogether.
Bars should not be 3D. This makes it more difficult to interpret the chart accurately because there are several potential ways to interpret the data points.
Use colors carefully. The rationale of using wildly different colors on a bar chart is to clearly differentiate the categories, but complementary colors (such as navy, royal, and sky blue) that vary in hue or saturation are just as easy to differentiate.

Related techniques

Bullet graph: Also known as a bullet graph, this is a variation on the bar graph that packs several dimensions of data into a bar. For instance, a bar showing revenue can show the overall quantitative scale, the performance measure itself, a comparative measure, and background fill colors representing a qualitative range.

Multiset bar chart: Also known as a clustered bar chart. A bar chart that shows multiple datasets by splitting them up to put relevant data points next to each other. For instance, a chart on employment figures could have a bar each for males and females. Use this to compare several datasets without having to switch back and forth between several charts. Datasets must be on the same scale and related to one another. An example is here: Percentage change in tuition and state appropriations.

Isometric bar chart: A variation on the multiset bar chart that uses 3D perspective to avoid breaking up data within the series by allowing overlapping bars. Use this to show several data series while retaining their original data structures. Because of its 3D requirements, it is very easy to produce chartjunk, so this visualization should be used with care.

Waterfall chart: A bar chart where each column's bar shows its contribution to a whole. Example: Recent state funding in perspective (click on chart to enlarge)

Navigation