How to find the frequency of the most frequent value (mode) of a series in polars?-CodePudding

import polars as pl

df = pl.DataFrame({
    "tags": ["a", "a", "a", "b", "c", "c", "c", "c", "d"] 
})

This is how to compute the most frequent element of the column using the .mode expression:

df.select([
    pl.col("tags").mode().alias("mode"),
])

How can I display also the frequency/count of that mode?

CodePudding user response：

There is a value_counts expression. This expression will return a Struct datatype where the first field is the unique value and the second field is the count of that value.

df.select([
    pl.col("tags").value_counts()
])

shape: (4, 1)
┌───────────┐
│ tags      │
│ ---       │
│ struct[2] │
╞═══════════╡
│ {"c",4}   │
├╌╌╌╌╌╌╌╌╌╌╌┤
│ {"a",3}   │
├╌╌╌╌╌╌╌╌╌╌╌┤
│ {"b",1}   │
├╌╌╌╌╌╌╌╌╌╌╌┤
│ {"d",1}   │
└───────────┘

Or if you want to have that result as a DataFrame:

(df.select([
    pl.col("tags").value_counts()
]).to_series().struct.to_frame())

shape: (4, 2)
┌──────┬────────┐
│ tags ┆ counts │
│ ---  ┆ ---    │
│ str  ┆ u32    │
╞══════╪════════╡
│ c    ┆ 4      │
├╌╌╌╌╌╌┼╌╌╌╌╌╌╌╌┤
│ a    ┆ 3      │
├╌╌╌╌╌╌┼╌╌╌╌╌╌╌╌┤
│ d    ┆ 1      │
├╌╌╌╌╌╌┼╌╌╌╌╌╌╌╌┤
│ b    ┆ 1      │
└──────┴────────┘

Edited: Which can be even simpler:

df["tags"].value_counts()

CodePudding user response：

I found this method, but I'm not sure if there's a better way:

df.select([
    pl.col("tags").mode().alias("mode"),
    pl.col("tags").filter(pl.col("tags") == pl.col("tags").mode()).count().alias("count")
])

Output:

shape: (1, 2)
┌──────┬───────┐
│ mode ┆ count │
│ ---  ┆ ---   │
│ str  ┆ u32   │
╞══════╪═══════╡
│ c    ┆ 4     │
└──────┴───────┘