CSV Anomaly Detector

CSV Anomaly Detector is a free online tool that works entirely in your browser — no signup, no data uploaded, no limits.

Some Wrong Numbers Are Obvious. Most Are Not.

A negative age or a salary of zero is easy to spot manually. The numbers that actually cause problems are the ones that look plausible at a glance but sit far outside the normal range for your dataset — an order of 15,000 dollars in a file where every other order is under 300, or a delivery time of 400 days where everything else is under 10. These hide in plain sight in a spreadsheet with hundreds of rows.

This tool applies the Interquartile Range method, a well-established statistical approach, to every numeric column in your data. It calculates where the middle 50% of your values sit and flags anything that falls unusually far outside that range as a potential anomaly.

Why IQR Rather Than a Fixed Threshold

A fixed rule like "flag anything over 1,000" only works if you already know what normal looks like for every column, and that knowledge does not transfer between datasets. IQR calculates the normal range from your actual data, so it automatically adapts whether you are analyzing order amounts in the hundreds or revenue figures in the millions.

Outliers Are Candidates, Not Verdicts

Flagging a row as an outlier does not mean it is wrong. It means it is statistically unusual and deserves a second look. Sometimes that review confirms a data entry mistake — an extra digit typed by accident. Sometimes it confirms a genuinely large but valid transaction. The tool's job is to point you at the right rows; deciding what to do about them is still a human judgment call.