ToolSphere

Scatter Plot Generator

More tools

Plot any two numeric columns from a CSV file against each other, with the Pearson correlation coefficient shown alongside.

Drag & drop your file here, or browse

Supports single file (.csv,text/csv)

Or paste CSV text directly

Upload or paste a CSV file to begin.

On this browser only

Give feedback

Did this tool help?

Tell us what worked, what didn’t, or what you’d like next.

Opens Google Forms in a new tab with this tool’s name filled in. No sign-in or email required. Feedback is optional.

About Scatter Plot Generator

Scatter Plot Generator

Plots any two numeric columns against each other as a scatter plot, with the exact Pearson correlation coefficient computed and displayed alongside — letting you both see the relationship visually and confirm it numerically in one view.

A scatter plot can reveal patterns a correlation coefficient alone hides, such as a clear non-linear relationship or distinct clusters, both of which can coexist with a near-zero linear correlation value.

What the plot shows that the number cannot

Anscombe's quartet is the standard demonstration: four datasets with identical means, variances and correlation coefficients that look completely different when plotted. One is a clean linear relationship, one is a curve, one is a perfect line with a single outlier, and one is a vertical stack with one stray point. Reporting r alone would call them the same dataset.

So the two views answer different questions. The coefficient tells you how strong a linear relationship is. The plot tells you whether a linear relationship is the right thing to be measuring.

Patterns worth recognising

A curve. Correlation near zero with an obvious U shape means the variables are strongly related, just not linearly. Fit a transformed term or a non-linear model rather than concluding there is no relationship.

Fanning out. If the spread of y widens as x grows, the variance is not constant. Linear regression's confidence intervals assume it is, so they will be wrong even though the fit looks reasonable.

Separate clusters. Two distinct clouds usually mean a hidden grouping variable — two products, two regions, two time periods — and the correlation across the combined data may not exist within either group.

One point far from the rest. A single outlier can create or destroy a correlation on its own. Identify the row before deciding whether it is an error or the most interesting data point you have.

Correlation and causation

The warning is worn out but the mechanism is worth stating: two variables can move together because one causes the other, because both are driven by a third, or because of how the sample was selected. Ice-cream sales and drowning deaths correlate strongly, and temperature explains both. A scatter plot never distinguishes these — only the design of the study does.

Guides and tutorials

Frequently asked questions

Why look at a scatter plot if I already have the correlation coefficient?

Correlation only captures linear relationship strength — a scatter plot can reveal non-linear patterns, clusters, or outliers that a single correlation number doesn't show, even when that number is close to zero.

Can I choose the same column for both axes?

Yes, though the result is just a straight diagonal line (perfect correlation with itself) — useful mainly as a sanity check, not a meaningful analysis.

What does a curved scatter plot with near-zero correlation mean?

That the variables are strongly related but not linearly. Pearson's r only measures straight-line association, so a clean U shape can score near zero. Transform a variable or fit a non-linear model.

Why does my correlation change so much when I remove one point?

Because a single outlier can create or destroy a linear relationship on its own, especially in small datasets. Find the row and decide whether it is an error or a genuine extreme before dropping it.

What do two separate clusters in the plot mean?

Usually a hidden grouping variable — two products, regions or time periods mixed together. The correlation across the combined data may not hold within either group, which is Simpson's paradox in practice.

Have a suggestion for Scatter Plot Generator?Share your feedback ↗