Samples drop from the top of the upper panel under simple gravity and stack into bins by value, building a live histogram — the same dataset the 3D version renders as a beeswarm cloud, but here order statistics are shown two ways instead of one camera angle: a physical stacking histogram, and below it a sorted rank curve (the empirical quantile function) plotting value against rank/n.
IQR = Q3 − Q1
Lower fence = Q1 − k · IQR
Upper fence = Q3 + k · IQR
Outlier ⇔ value < lower fence OR value > upper fence
- Sample size / outlier rate — regenerates the underlying dataset and replays the drop.
- Fence multiplier k — drag the orange lines on the histogram, or the slider; both stay in sync and re-classify the same dataset without a new drop.
- The rank-plot panel marks rank/n = 0.25, 0.50, 0.75 with dotted guides at the exact x-position where Q1, median and Q3 sit by definition.
Real-world relevance: this exact 1.5×IQR rule is the default outlier rule in pandas' .describe()-adjacent tooling and most BI box-plot widgets — often the first, cheapest anomaly filter run before a heavier ML-based detector.