[FEATURE] interquartile range (IQR)

[bennycode]

To calculate the average of a dataset while removing big outliers, the interquartile range (IQR) method can be used.

The IQR serves as a robust measure of variability because it is less sensitive to outliers compared to the range.

Resources: https://www.youtube.com/watch?v=qLYYHWYr8xI&t=106s

• A small IQR (like 3) indicates that the middle 50% of the data is relatively close together, showing lower variability in the core values.
• A larger IQR would indicate that the data is more spread out, suggesting greater variability within the central 50%.
• The IQR can be used to compare variability between different datasets. For example, if another dataset has an IQR of 8, its middle 50% is more spread out than yours.

[bennycode]

Sample values

From: https://www.khanacademy.org/math/cc-sixth-grade-math/cc-6th-data-statistics/cc-6th/v/calculating-interquartile-range-iqr

---

4
4
6
7
10
11
12
14
15

media = 10
iqr = 13-5 = 8

---

7
9
9
10
10
10
11
12
12
14

median = [10,10]
iqr = 12-9 = 3

---

More here: https://en.wikipedia.org/wiki/Interquartile_range

[bennycode]

Other Recommendations

• For mild outliers, use the trimmed mean or Winsorized mean
• For highly skewed datasets with severe outliers, the median is the most robust choice
• If the data has positive values and spans orders of magnitude, the geometric mean is ideal