Outlier Detection Tool
Flag values outside Tukey's fences, beyond three standard deviations, or beyond a modified z of 3.5.
What this page computes
Paste one number per line. Three published rules run on the same column: Tukey's fences at Q1 − 1.5·IQR and Q3 + 1.5·IQR with type 7 quartiles; the classical z = (x − x̄)/s with |z| above 3, where s uses the n − 1 divisor; and the modified z = 0.6745(x − median)/MAD with |z| above 3.5, where the constant is the standard normal 75th percentile and MAD is the median absolute deviation. Every flagged value is listed with its line number. Figures print to six significant figures.
An outlier is a value a rule flags, not a mistake, and the page does not delete or replace anything. The three rules disagree by design, which is why all three are shown. A column with zero variance or a zero MAD is refused rather than divided into.
All three rules run in the tab, so the same column flags the same lines every time.
The page applies three published outlier rules to one column and names what each one flags: Tukey fences at Q1 − 1.5·IQR and Q3 + 1.5·IQR, the classical z = (x − x̄)/s with |z| above 3, and the modified z = 0.6745(x − median)/MAD with |z| above 3.5. For 2, 4, 4, 4, 5, 5, 7, 9 only the modified rule flags the 9, at 6.0705, while the other two flag nothing. Three rules are shown side by side because they disagree, and the page never removes a value.
Flag possible outliers by three standard rules at once. Every flagged value keeps its line number and its statistic.
How to use it
Paste your data with one value per line; a comma inside a line is refused because this page reads a column. Each flagged value is printed with the line it came from and the statistic that flagged it, so nothing is silently deleted and the original paste stays authoritative.
Why the three rules disagree
The three rules have different assumptions and it shows: Tukey fences need only the quartiles, the classical z uses the mean and standard deviation which a single extreme value inflates, and the modified z uses the median and the median absolute deviation which it does not. The constant 0.6745 in the modified rule is the standard-normal 75th percentile, from Iglewicz and Hoaglin’s monograph. A flag is a prompt to look, not a verdict, and the page says so.
What to Expect
Best for
- Checking a column for values worth a second look
What this tool keeps
- Nothing leaves this tab: the arithmetic runs in the page you are reading, and a reload discards both the input and the result.
Sample input
2 4 4 4 5 5 7 9
Sample output
n = 8, median = 4.5, IQR = 1.5 fences: outside 1.75 … 7.75 (Q1 − 1.5·IQR, Q3 + 1.5·IQR) Tukey fences: 9 (line 8) classical |z| > 3: none modified |z| > 3.5, MAD = 0.5: 9 (line 8), modified z = 6.0705
Frequently asked questions
Which rule should I trust?
None alone. The page shows all three because they disagree — Tukey fences, the classical |z| above 3 and the modified |z| above 3.5 rest on different assumptions — and a flag is a prompt to inspect the value, not a verdict.
Does the page remove the outliers it finds?
No. It lists each flagged value with its line number and the statistic that flagged it, and leaves the data exactly as pasted.
Is my data uploaded?
No. All three rules are evaluated in this page, so the column never leaves the tab and nothing is stored.
Build custom tools for your store
Join the SDK program to create useful ecommerce utilities for your store.