Data Profiler

Look at the data before you trust it

Every dataset arrives with a story about what it contains, and the story is usually optimistic. A column that should be a number is forty percent empty. An identifier that should be unique repeats. A category that should have six values has nineteen, because three of them are spelling variants. Profiling is the ten second version of finding all of that out.

Missing values change every calculation

An average over a column that is a third empty is an average of two thirds of your data. Knowing the fill rate per column tells you which numbers you can quote and which need a caveat, and it is the first thing worth checking after a merge or an export.

Distinct counts reveal the shape

A column with two distinct values is a flag. One with as many distinct values as rows is an identifier. One with a handful is a category worth grouping by. The distinct count usually tells you more about how to use a column than its name does.

Distribution catches the outliers

Mean and median sitting far apart is the signature of a skewed column or a stray value with too many zeros. The distribution strip makes that visible without plotting anything, which is often enough to decide whether to investigate.