[Python] How to create a cross table with percentages in Polars?

I would like to create a cross table that shows, in each cell, the percentages of rows over the total number of rows.

Inspired by this post I started with:

df = pl.DataFrame({"a": [2, 0, 1, 0, 0, 0], "b": [1, 1, 1, 0, 0, 1]})

crosstab = (
df.pivot(on="b", index="a", values="b", aggregate_function="len", sort_columns=True)
.fill_null(0)
.sort("a")
)

crosstab


and then inspired by polars' user guide I tried to convert values into percentages with:

def perc_cols(df):
tot = df.select(~cs.by_index(0)).to_numpy().sum()
for col in df.columns[1:]:
yield ((pl.col(col) / tot) * 100)

crosstab.select(cs.by_index(0), perc_cols(crosstab))


but I get an error:

TypeError: cannot create expression literal for value of type generator.


notice that both crosstab.select(cs.by_index(0)) and crosstab.select(perc_cols(crosstab)) works as expected.

Continue reading...