Radar charts explained: when to use one, examples, and how to make one

A radar chart compares several measurements by placing them on spokes around a circle. Each subject gets a connected profile, making it easier to see where products, teams, or other groups differ across the same criteria.

For example, one product might score highly on reliability and support but lower on ease of use. A radar chart keeps those differences visible together, although aligned bars are easier to read when the gaps are small.
We'll walk through how to read a radar chart, where we'd use one, and when we'd choose bars instead. The real-data examples show both sides of that decision.

What is a radar chart?

A radar chart, also called a spider chart or spider web chart, compares several numerical measurements for one or more subjects. Each spoke represents a measurement. A point's distance from the center shows its value on that spoke, and a line connects the points belonging to the same subject.

For a product comparison, the spokes might represent ratings for reliability, support, and ease of use. Each product gets its own line. Before interpreting the outline, we need to know who assigned those ratings and what the scale means. A score of 80 tells us very little on its own.

Radar charts work best as a compact view of a profile. They do not automatically produce an overall score, and a larger filled shape does not establish which subject is best.

Radar chart key components & parts

A radar chart contains several parts that work together:

• Center baseline: The starting value where the spokes meet, usually zero. If the scale begins elsewhere, that starting point needs to be explicit.
• Radial axes: The spokes, each representing a different measurement, such as fuel economy, horsepower, or weight.
• Scale and gridlines: The labeled intervals used to read values. Concentric circles or polygons help readers judge distance from the center.
• Data points and plotted polygon: Each point marks a subject’s value on one spoke. Connecting those points forms its profile; a light fill can make the shape easier to follow.
• Legend: Identifies the subjects represented by different colors or line styles.

Distance along a spoke carries the value. The filled polygon helps us follow a profile, but its area is not an overall score.

How to read a radar chart

Start with one spoke, then compare the subjects on that measurement before moving around the chart.

1. Read the spoke labels. Establish what is being measured and whether the values describe an individual, a group average, or a calculated score.
2. Check the scales. Find the starting value, outer limit, units, and direction. Equal distances from the center are comparable only when the scales support that comparison.
3. Follow one line at a time. Use the legend to identify the subject, then look for the measurements where it extends farther outward or falls closer to the center.

Farther out means a larger plotted value. Whether that is desirable depends on the measure. A longer battery life may be useful; a longer loading time usually isn't.

Axes and scales

Each spoke has a numerical scale, with smaller values near the center and larger values toward the edge. The scale determines where each point sits and therefore affects the shape of the chart.

Measurements in different units, such as price in dollars and battery life in hours, need a common basis for comparison. One option is to convert them to scores from 0 to 100. If 0 represents the lowest observed value and 100 the highest, those scores show position within the sample, not product quality. A higher price is not necessarily better, while longer battery life may be.

Some data already uses a shared scoring system. In the decathlon example below, every event result is expressed in points, so all spokes use the same points range.

For comparisons over time, the scale’s reference values must stay fixed. Recalculating the endpoints each month could change a product’s plotted score even when its actual measurement stays unchanged. Separate bar charts are a clearer option when retaining the original units matters more than comparing overall profiles.

Filled versus line radar charts

An outline with visible markers usually gives us enough structure to follow each subject. Light shading can help identify a single profile, but overlapping fills can obscure the lines we need to compare.

Filled area also depends on the order of the spokes. The example below uses the USA vehicle profile from the Auto MPG dataset, with five group averages converted to fixed 0–100 scales. Both arrangements use exactly the same values and scale limits.

Switch between Order A and Order B. Only the spoke order changes, yet the outline and enclosed area change too. That area cannot serve as a dependable total of the measurements.

We would keep spoke order consistent across every radar chart in a comparison. Otherwise, readers have to relearn the shape each time.

When to use a radar chart

We would consider a radar chart when the profile itself helps answer the question and the reader can follow a small number of subjects without repeatedly checking the legend.

• Skills assessments: Compare current and target ratings across defined skills. Explain the assessment rubric and who supplies the ratings, especially when they are self-reported.
• Product scoring: Compare products evaluated against the same criteria. Keep the scoring rules consistent and make clear whether a larger value always means a better result.
• Benchmark profiles: Show where competitors perform differently across a fixed set of tests. A common scoring system helps, but the individual results still need to be available.

The useful question is usually “Where do these profiles differ?” If the decision depends on the exact size of each difference, we'd move to bars.

When not to use a radar chart

Our default for a precise comparison is a set of small bar charts, one per measure. Each panel puts the subjects on a shared baseline, so readers can compare lengths without following spokes around a circle.

We would choose that alternative in these situations:

You can compare individual measures with bar graphs, using one panel for each measurement. When several panels use the same measure and units, keep their scales consistent. If the panels contain different units, label each scale and avoid comparing bar lengths across panels.

For monthly or yearly measurements, show the trend with a line graph. A radar chart connects the last spoke back to the first, which can imply a cycle that the data do not have.

We would also avoid treating the number of spokes as a checklist to fill. Every extra measure adds another label and another change of direction. Include a measure because it affects the comparison, not because the circle looks sparse.

For other questions, our guide to types of graphs and charts helps narrow down the alternatives.

Radar chart examples with real data

Vehicle profiles: power, size, and fuel economy

The Auto MPG dataset contains historical vehicle records from model years 1970 to 1982. We compare average fuel economy, horsepower, weight, engine displacement, and acceleration time across its USA, Europe, and Japan origin groups.

After excluding six records with missing horsepower, the comparison contains 392 vehicles: 245 in the USA group, 68 in Europe, and 79 in Japan.

These measurements use different units, so we convert each to a fixed 0–100 scale. Zero represents the lowest observed value for that measurement among the 392 vehicles; 100 represents the highest. We apply those same reference values to every group. Hover details retain the original averages and units.

The USA profile extends farther on horsepower, weight, and engine displacement. The Japan profile extends farther on fuel economy, averaging about 30.5 mpg compared with 20.0 mpg for the USA group. The colored areas make those contrasting profiles visible without collapsing them into one rating.

Farther outward does not always mean better. On the acceleration-time spoke, it means more seconds. These are historical group averages, and they hide differences between individual vehicles and model years; they should not be read as ratings of today’s vehicles.

Decathlon scores: different events on a common points scale

The official Athens 2004 decathlon results give us a different kind of profile. We compare Roman Šebrle and Bryan Clay using their published points for each event. Those points already account for the event's scoring rules, so we can compare the profiles without putting seconds and meters on the same radial scale.

Clay leads in the 100 meters, long jump, and discus; Šebrle leads in the other seven events. The profiles overlap closely, making this a useful place to question our chart choice. Ten spokes are a lot to follow when several gaps are small.

The same scores as small bar charts

Here we give each event its own panel and keep the points scale identical throughout.

The event-by-event differences are easier to inspect. Clay's largest advantage is in the 100 meters; Šebrle's largest gains are in shot put and high jump. Šebrle finishes 73 points ahead overall, a result we calculate by adding the event scores. The radar's enclosed area is not part of that calculation.

We’d use these small bar charts when readers need to compare each event’s points and see where the overall score difference comes from.

Radar, spider, star, web, and polar charts: what’s the difference?

Several names describe the same spoke-based chart. The useful differences concern how profiles are arranged and whether the angle around the circle carries information.

Spider chart

A spider chart is another name for a radar chart. The name describes the spoke-like axes and connected outline. Our decathlon comparison is both a radar chart and a spider chart: each event occupies a spoke, and each athlete’s points form a profile.

Web chart or cobweb chart

Web chart, spider web chart, and cobweb chart also commonly refer to radar charts. The names emphasize the grid connecting the spokes. For example, a product scorecard with five criteria rated from 1 to 5 may be called a web chart whether its profiles use outlines or light fills. The name does not introduce a different calculation.

Star chart or star plot

Star chart is often used as another name for a radar chart. In statistical graphics, star plots commonly place each observation in its own small chart, making it possible to compare shapes without overlapping the profiles.

Below, we separate the same three vehicle-group profiles into individual panels. The measurements, spoke order, and 0–100 scales stay fixed; only the layout changes.

Each panel is easier to follow on its own, although comparing close values across panels still takes more effort than reading aligned bars.

Kiviat diagram

A Kiviat diagram uses the same basic structure: metrics on spokes, with values connected into a profile. For example, a software team might plot test coverage, documentation coverage, and the percentage of requests meeting a response-time target. A target outline can show where results fall short, provided every metric’s scale and target are explained.

The name alone does not imply a particular scoring formula or require a target outline.

Radar chart vs. polar chart

A polar chart is a broader family of charts that positions values using distance from a center and angle around it. A radar chart uses that arrangement for different measurements placed on spokes.

The distinction becomes clearer with wind direction. In the illustration below, north, east, south, and west occupy meaningful compass positions, while bar length shows the percentage of observations from each direction. Rearranging those directions would change the meaning. On a radar chart, spoke positions are an ordering choice, although we should keep that order consistent when comparing profiles.

This illustration uses 100 hypothetical wind observations to demonstrate the coordinates; it does not represent a weather station’s measurements.

How to create a radar chart

Open Deepnote’s radar chart maker and upload your spreadsheet or paste a table with column headings. Identify the subjects, the measurements that belong on the spokes, and the units or scoring scale.

For data that still need preparation, we can use Deepnote Agent to calculate the summaries in a notebook and generate the chart in Python. That keeps the raw values, calculation, and visualization available together when someone asks how a point was produced.

Prepare the measurements

Use one value for each subject and spoke. If your data contain multiple observations per subject, choose a summary, such as the mean, and explain how you calculated it.

For this walkthrough, we’ll use the Iris dataset available through scikit-learn. Each row represents one flower. We’ll calculate each species’ mean sepal length, sepal width, petal length, and petal width. All four measurements use centimeters, so they can share a radial scale without conversion.

Check for missing values before calculating the means or drawing the chart. Replacing a missing measurement with zero would introduce a value that was never observed.

Describe the chart and its limits

Ask Deepnote Agent to load the Iris dataset from scikit-learn, calculate the species means, and create the radar chart.

Then enter:

Load the Iris dataset from scikit-learn. Create an interactive radar chart comparing mean sepal length, sepal width, petal length, and petal width for each species. Check the number of flowers per species before calculating the means. Keep that spoke order and use a shared radial scale from 0 to 7 centimeters. Use distinct colors, visible markers, and light transparent fills. Include the species name, measurement, and mean in the hover labels, and keep the calculation visible in the notebook.

For a scored comparison, include the scoring rule or supply the existing scores. Ask for any transformation to be shown in the notebook so you can check how the plotted values relate to the source.

Compare, then share

Before settling on the radar chart, ask Agent to show the same values as small bar charts. Check which view makes the important differences easier to explain. In our decathlon example, that second view carries the detailed comparison more clearly.

You can share or embed the chart output while keeping the calculations in the notebook for your team to review. If you're still deciding which format fits your data, start with the AI chart generator and describe the comparison you need to make.