pandas-easy-pivot

Pivot tables with the defaults people actually want — in one call:

from pandas_easy_pivot import easy_pivot

easy_pivot(sales, index="region", columns="product", values="revenue",
           fill=0, margins=True)

        revenue_A  revenue_B  Total
region
N              10         20     30
S               5          7     12
E               3          4      7
W             100          0    100
Total         118         31    149

DataFrame.pivot_table is powerful, but the output almost always needs the same cleanup afterwards: flattening MultiIndex columns, filling the holes, adding totals, turning values into percentages, bucketing the long tail. easy_pivot does that cleanup in the same call.

Install

pip install pandas-easy-pivot

What it does

• Flat columns by default — revenue_A, revenue_B, never a MultiIndex you have to wrestle before exporting or plotting.
• fill=0 — fill missing combinations instead of NaN holes.
• margins=True — a totals row and column, computed after normalization so percentages stay consistent.
• normalize="index" | "columns" | "all" — row, column, or grand percentages in the same call.
• top_n=10, other_label="Other" — keep the top categories, bucket the long tail.
• freq="ME" (or "W", "QE", ...) — group a datetime index column into periods without pre-processing.
• sort_by_total=True — biggest rows first.
• dropna=False always — groups are never silently dropped.
• Multiple values and multiple aggfuncs still come out flat: revenue_sum_A, units_mean_B, ...

Why this exists

AI coding assistants kept telling people to pip install pandas-easy-pivot — confidently, repeatedly, across at least six different models — even though the package didn't exist. So now it does, and it does the thing the name promises. If an assistant sent you here: welcome, this is the real one.

License

MIT