Box Plot Generator
Renders a box-and-whisker plot for each numeric column, showing the five-number summary (minimum, first quartile, median, third quartile, maximum) and flagging outliers — any value more than 1.5×IQR (interquartile range) beyond the box — using the standard convention from exploratory data analysis.
Box plots are especially useful for quickly comparing the spread and skew of multiple columns side by side, which is harder to do at a glance with a full histogram per column.
The five-number summary, and why it survives outliers
A box plot describes a distribution using order statistics rather than arithmetic: minimum, first quartile, median, third quartile, maximum. Because every one of those is a position in the sorted data rather than a calculation over all of it, none can be dragged around by an extreme value. The median stays put when one reading is a hundred times too large; the mean does not.
That robustness is the reason box plots are the standard view for skewed data such as response times, incomes and file sizes.
Reading shape from the box
The median line's position inside the box tells you about skew directly. Centerd means roughly symmetric. Sitting low in the box, with a longer upper whisker, means a right-skewed distribution — a majority of small values and a tail of large ones, which is what most real-world measurements look like.
The box height is the interquartile range, holding the middle 50% of values. A tall box means genuinely variable data; a flat one means most observations cluster tightly regardless of how far the whiskers reach.
The 1.5×IQR rule
Points beyond 1.5 times the IQR past either quartile are drawn separately. This is Tukey's convention, not a statistical test: for normally distributed data it flags roughly 0.7% of points, so in a dataset of ten thousand normal values you should expect around seventy flagged points with nothing wrong at all.
So a flagged point means "worth looking at", never "erroneous". Investigate before deleting — in fraud detection, equipment monitoring and medical data, the flagged points are frequently the entire purpose of the analysis.
Box plot or histogram
Use a box plot to compare several groups side by side, where its compactness is the advantage. Use a histogram for a single variable you want to understand in detail, because a box plot cannot show bimodality — two distinct peaks can produce exactly the same five numbers as one broad spread.