Plot relationships with a scatter plot.

How to use this scatter plot tool

This free scatter plot maker turns paired numeric observations into an editable chart. Use it to inspect relationships between variables, compare groups, identify clusters, and find values that deserve closer attention.

1. Import paired numeric data

Select Import data and upload a CSV, TSV, TXT, JSON, XLSX, or XLS file, or paste a table directly. Choose one numeric column for the X axis and another for the Y axis. Additional numeric columns can be plotted as separate Y series against the same X values.

2. Review and edit each point

The editable table keeps the source precision and updates the plot as values change. Add or remove observations, reverse or sort rows, add another series, or reset the sample data at any time.

3. Style and explore the plot

Rename each series, choose its color, toggle points and grid lines, and add descriptive axis labels. Zoom and pan through dense datasets while the loading overlay keeps large updates responsive.

4. Export the finished chart

Download a PNG for documents and presentations or an SVG for scalable layouts. Export size, aspect ratio, render scale, background color, and transparency are configurable.

How to read a scatter plot

Each point combines one X value and one Y value. An upward pattern may indicate a positive association, a downward pattern may indicate a negative association, and a diffuse cloud may indicate little visible relationship. A scatter plot shows association, not causation.

Look for shape, strength, and unusual points

Consider whether the relationship is linear or curved, how tightly points follow the pattern, and whether separate clusters are present. Outliers can be meaningful observations or data-quality problems, so investigate them before removing them.

Use honest scales and clear units

Label both axes with the measured variables and units. Keep enough range to show the full dataset, and avoid scale choices that exaggerate weak patterns. When comparing series, use consistent measurements so their positions remain interpretable.

Practical example: compare response time and satisfaction

Place response time on the X axis and satisfaction score on the Y axis, with one point per support case or summarized group. The cloud may show whether longer waits tend to accompany lower scores, whether the relationship changes after a threshold, or whether separate customer groups follow different patterns.

Do not rely on a coefficient alone

A correlation value summarizes one aspect of association and can hide curves, clusters, unequal spread, and influential outliers. Inspect the points first, identify how repeated measurements were handled, and consider sample size and uncertainty. For many variables, use a correlation matrix to screen pairs, then return to individual scatter plots for interpretation.

Overplotting can conceal frequency

When many points share similar coordinates, transparency and smaller markers help, but they may still hide density. A hexbin chart groups nearby observations into counts and is often clearer for large datasets. Keep the raw scatter view available when individual outliers or labels matter.

Document the observational unit

State what each point represents, the measurement period, filters, units, and any transformations. Points from the same person or location may not be independent. Association in an observational plot should not be presented as evidence that changing X will cause Y to change.