MiscKit

Scatter Plot Maker

Paste two columns and see whether they move together. Add more columns for more series, then export a PNG.

About this tool

A scatter plot is the fastest way to find out whether two measurements are related. Ad spend against signups, temperature against energy use, study time against test scores, dose against response — plot them as points and the answer is usually visible in a second: a rising cloud, a falling one, a shapeless blob, or a curve that no straight line would fit.

It is also good at showing what a correlation number hides. Two datasets can share the same average, the same spread and the same correlation while looking nothing alike — one an even band of points, the other a tight curve with a single far-off outlier dragging the statistic around. Only the picture tells you which one you have, which is why plotting before summarising is worth the extra minute.

Both axes are scaled to round numbers that cover your data with a little room at the edges, so points never sit on the frame. Paste straight from a spreadsheet: tabs, commas and runs of spaces all work, a header row is detected and skipped, and thousands separators and currency symbols are stripped before the numbers are read.

Frequently asked questions

What should each column contain?

Two columns: X first, then Y. Every row is one point, so "2.5, 58" plots a dot at x = 2.5 and y = 58. If the first column is not numeric — month names, product codes — it cannot be used as an axis, so points are spread evenly along X by row order instead. That still shows the shape of the Y values, but the horizontal spacing no longer means anything.

Can I plot more than one series?

Yes. Add extra numeric columns after the X column and each one becomes its own colour, sharing the same X values — useful for comparing two measurements taken under the same conditions. A legend appears automatically once there is more than one series.

Does a pattern in the chart prove one thing causes the other?

No, and this is the mistake scatter plots invite most often. A rising cloud of points means the two measurements move together, nothing more. Both may be driven by a third factor you did not record, or the relationship may run the opposite way from what you assume. The chart is where you start looking, not where you conclude.

What if some values are missing?

Empty cells are read as missing rather than as zero, so they simply produce no point. This matters: turning a blank into 0 would drag your trend toward the origin and invent a relationship that is not in the data.

Is anything uploaded?

No. Parsing and drawing both happen in this page, and no request goes out while you work. The PNG export runs through the browser Canvas API at 2x resolution, so measurements you are not free to share never leave your machine.

Other tools