Outlier Detection Explanation Calculator
Use our free Outlier detection explanation tool to get instant, accurate results. Powered by proven algorithms with clear explanations.
Reviewed by Daniel Agrici, Founder & Lead Developer
Formula
IQR: Outlier if x < Q1 - 1.5*IQR or x > Q3 + 1.5*IQR
The IQR method defines outliers as points beyond 1.5 times the interquartile range from the quartiles. The Z-Score method flags points more than a threshold number of standard deviations from the mean. The MAD method uses the median absolute deviation scaled by 1.4826 for robustness against the outliers themselves.
Worked Examples
Example 1: Temperature Sensor Readings
Problem:Readings: 22, 23, 21, 22, 85, 23, 22, 21, 23, 22. Detect outliers using IQR method (1.5x).
Solution:Sorted: 21, 21, 22, 22, 22, 22, 23, 23, 23, 85\nQ1=22, Q3=23, IQR=1\nLower fence: 22 - 1.5(1) = 20.5\nUpper fence: 23 + 1.5(1) = 24.5\nOutlier: 85 (sensor malfunction)\nClean mean: 22.1 vs raw mean: 28.4
Result:1 outlier detected (85) | Clean mean: 22.1 | Raw mean: 28.4
Example 2: Student Test Scores
Problem:Scores: 72, 75, 80, 78, 82, 76, 5, 79, 81, 77. Z-Score method with threshold 2.
Solution:Mean: 70.5, StdDev: 22.1\nZ-score of 5: (5-70.5)/22.1 = -2.96\nAll others: z-scores between -0.7 and 0.8\nOnly value 5 exceeds |z| > 2\nLikely a student who missed most of the exam
Result:1 outlier detected (5) | Z-score: -2.96 | Clean mean: 77.8
Frequently Asked Questions
What is an outlier and why should I detect them?
An outlier is a data point that significantly differs from the rest of the dataset. Outliers can be caused by measurement errors, data entry mistakes, natural variation, or genuinely unusual observations. Detecting outliers is important because they can skew statistical analyses: a single extreme value can shift the mean dramatically, inflate standard deviation, and distort regression models. For example, in a dataset of salaries [40K, 45K, 50K, 55K, 2M], the mean is ~438K which misrepresents the typical salary. Outlier detection helps you decide whether to investigate, remove, or separately analyze these extreme values.
How do I choose the right threshold for outlier detection?
The threshold determines the sensitivity of outlier detection. For the IQR method, 1.5x IQR is the standard for mild outliers (used in box plots) and 3.0x IQR for extreme outliers. For Z-Score, a threshold of 2 flags about 5% of normally distributed data (more aggressive), while 3 flags about 0.3% (more conservative). For MAD, a threshold of 3 is standard. Start with default thresholds and adjust based on your domain knowledge: in financial data where extreme values are common, use higher thresholds; in manufacturing quality control where precision matters, use lower thresholds. Always investigate flagged outliers before removing them.
References
Reviewed by Daniel Agrici, Founder & Lead Developer · Editorial policy