You plotted a histogram of a real column. Now what? Read five realistic columns the way a practitioner does: name the shape, infer the process that generated it, and predict what it will break downstream.
Sign in to save progress
The histogram is your first interview with the data
Before any model, a histogram answers one question: what process generated this column? Skew points to multiplicative processes. Twin peaks mean two populations are hiding in one column. A tail that refuses to fade means rare extreme events dominate your averages. An isolated spike is usually a bug wearing a data costume. And a wall of zeros means two different behaviors got mixed. The mathematics of the named distributions behind these shapes lives in the Data Distributions statistics guide; this one is about reading shapes in the wild and knowing the consequences.
How this playground stays honest: all five columns are generated once with seeded generators (400 rows each), so they are identical on every visit. Every histogram, mean, median, percentile, and skewness value on this page is computed live from those rows. Nothing is faked.
order_amount · n = 400 · USD
One month of e-commerce order values. Most orders are small; a minority are large.
Diagnosis
Read the histogram first. What shape do you see, and what process could have generated it? Then run the diagnosis to check your read against the computed statistics.
Your Mission
Shape-Matching Exercise
Four fresh columns, no context, no mean or median lines. Diagnose each from shape alone. Get at least 3 of 4 right.
Mystery column 1
Mystery column 2
Mystery column 3
Mystery column 4