How to apply value_counts() to multiple columns in polars python?

Question

I am trying to apply a simple value_counts() to multiple columns on a dataframe in polars but getting error.

import polars as pl
import pandas as pd

data:

sample_df = pl.DataFrame({'sub-category': ['tv','mobile','tv','wm','micro','wm'],
              'category': ['electronics','mobile','electronics','electronics','kitchen','electronics']})

Failed Attempts:

#1
sample_df.apply(value_counts())

#2
sample_df.apply(lambda x: x.value_counts())

#3
sample_df.apply(lambda x: x.to_series().value_counts())

#4
sample_df.select(pl.col(['sub-category','category'])).apply(lambda x: x.value_counts())

#5
sample_df.select(pl.col(['sub-category','category'])).apply(lambda x: x.to_series().value_counts())

But if I convert it to Pandas dataframe then it works:

sample_df.to_pandas().apply(lambda x: x.value_counts())

Answer 1

You could reshape with .melt()

shape: (12, 2)
┌──────────────┬─────────────┐
│ variable     ┆ value       │
│ ---          ┆ ---         │
│ str          ┆ str         │
╞══════════════╪═════════════╡
│ sub-category ┆ tv          │
│ sub-category ┆ mobile      │
│ sub-category ┆ tv          │
│ sub-category ┆ wm          │
│ sub-category ┆ micro       │
│ …            ┆ …           │
│ category     ┆ mobile      │
│ category     ┆ electronics │
│ category     ┆ electronics │
│ category     ┆ kitchen     │
│ category     ┆ electronics │
└──────────────┴─────────────┘

In which case it is then the length of each group:

df.melt().group_by(pl.all()).len()

shape: (7, 3)
┌──────────────┬─────────────┬─────┐
│ variable     ┆ value       ┆ len │
│ ---          ┆ ---         ┆ --- │
│ str          ┆ str         ┆ u32 │
╞══════════════╪═════════════╪═════╡
│ category     ┆ kitchen     ┆ 1   │
│ sub-category ┆ tv          ┆ 2   │
│ sub-category ┆ mobile      ┆ 1   │
│ category     ┆ mobile      ┆ 1   │
│ sub-category ┆ wm          ┆ 2   │
│ sub-category ┆ micro       ┆ 1   │
│ category     ┆ electronics ┆ 4   │
└──────────────┴─────────────┴─────┘

.pivot() can be used to reshape into individual columns if required.

(df.melt()
   .pivot(
      index = "value",
      columns = "variable",
      values = "value",
      aggregate_function = pl.len()
   )
)

shape: (6, 3)
┌─────────────┬──────────────┬──────────┐
│ value       ┆ sub-category ┆ category │
│ ---         ┆ ---          ┆ ---      │
│ str         ┆ u32          ┆ u32      │
╞═════════════╪══════════════╪══════════╡
│ tv          ┆ 2            ┆ null     │
│ mobile      ┆ 1            ┆ 1        │
│ wm          ┆ 2            ┆ null     │
│ micro       ┆ 1            ┆ null     │
│ electronics ┆ null         ┆ 4        │
│ kitchen     ┆ null         ┆ 1        │
└─────────────┴──────────────┴──────────┘