.describe() for series of objects by takluyver · Pull Request #241 · pandas-dev/pandas (original) (raw)

See #210.

I don't know if series.dtype == object is the best way to check for non-numeric data, but it appears to work.

At present, I've used fields 'count', 'unique', 'top' (i.e. most common) and 'freq' (i.e. count of the most common). These seem like obvious things for a summary.

I also noticed a problem doing df.describe() if df has integer columns, so I copied some code from series to fix that.