Pie and donut charts display portions of a whole unit so users can compare data points from a total set. A donut chart also includes a summary metric in the center of the chart.
The objective of data visualization is for users to be able to quickly and clearly derive meaning from a set of data. Data visualization supports four common objectives:
Identifying trends: Users want to understand how a metric or set of metrics is changing over time, particularly if the changes correspond with other events in the user's services.
Identify aberrations/anomalies: Users want to spot deviations from a normal or expected range, often the significant increase or decrease of a particular metric.
Comparison: Users want to identify commonalities or divergences between two or more metrics.
Investigate problems: Users have been notified of a problem or found one during a review or investigation, and use a visualization to better understand what happened.
A data visualization can focus on one or more of these, as well as typically fulfill several of the these user objectives. It’s important to understand the primary and secondary needs of your users, and to choose the right data and visualization to meet those needs.
Mixed line and bar chart
Numeric or categorical
Continuous or snapshot
Number of metrics
1 to 8 data series
1 to 8 data series
1 to 4 data series
2 to 8 data points
1 to 5 data points
Data can generally be divided in two main types: numeric (any quantitative measure) and categorical (qualitative data, usually expressed with text labels). Identify the data type your user needs and select the chart that best supports the data.
Individual units of measurement, typically arrayed over a time or date range.
For example: A bar chart that shows the number of errors logged by an application over the past three months.
Qualitative delineation between sets of data. Categories can be a ranked or ordered series of data (such as high, medium, and low severity). Or they can be grouped by type, which has no standard order (such as different types of databases).
For example: A stacked bar chart that shows the total number and severity of alerts over the past seven days.
The time period of a chart is the way time is depicted in the chart's data and display.
Shows multiple data points over multiple points or periods of time.
For example: A line chart that shows CPU usage of an instance over the course of 7 days.
Shows metrics from a single point or period of time.
For example: A pie chart that shows the number of different alert types logged in a specific day.
Number of metrics
The number of metrics shown in a data visualization should be sufficient for the user to understand the visualization, but not too many that it becomes confusing or overwhelming.
There are two levels of metrics that should be considered.
ndividual numerical points of data charted in the visualization. On charts that use X and Y axis, such as line charts, a data point is plotted on a specific X and Y coordinate. On charts that use polar coordinates, such as pie charts, the data point is represented as a segment of the total area. For example: The temperature of a CPU at a specific time.
The number of data points shown should reflect the granularity users need to properly interpret the visualization.
Data points that are related to each other and grouped to form a series. Some charts, such as line or bar charts, might have multiple data series on the same visualization. Other types of charts, such as pie charts, have only a single data series.
For example: A set of CPU temperatures logged once a second for a minute.
The number of included data series shouldn’t clutter the chart and overwhelm the user. Refer to the guidance for each chart type for the number of data series to show.
Use appropriate meta information, such as titles, labels, and the legend, to describe the chart’s intention and ensure users understand how the data displayed relates to other information on the page.
Include the minimum number of metrics and information users need to complete their desired task. Refer to the guidelines for each chart type for specific guidance.
Avoid showing too many metrics on a single chart. If you’re over the recommended number of metrics, consider showing multiple charts or grouping metrics.
When showing a large number of metrics on a single chart, include data filters so users can decide which metrics to show.
Ensure chart placement and size fits within the visual hierarchy of the page. When using multiple charts at the same level of importance, they should have consistent sizing.
Minimize the number of charts displayed in a single view to avoid visual overload.
Use a consistent and accessible color palette for your data visualization. For more information, see data visualization colors for guidance.
Don't use data visualization for decoration.
Never display data in a way that could mislead users to a false conclusion. For example, if the Y axis of a line or bar chart does not start at 0, changes are exaggerated, potentially creating a misleading impression of significant changes over time.
Refer to the writing guidelines for different types of Cloudscape charts.