first_column_as_header in transpose

[stevenlis]

Problem description

The method transpose could be very handy when you have to review a list of variables of ids. I already find it's much easier to scroll vertically than horizontally.

import polars as pl

df = pl.DataFrame(
    {"id": ['a', 'b', 'c'],
     "col1": [1, 3, 2],
     "col2": [3, 4, 6]}
)

Right now, you can do the following but the first row is redundant.

df.transpose(
    include_header=True,
    header_name='id',
    column_names=df['id']
)

shape: (3, 4)
┌──────┬─────┬─────┬─────┐
│ id   ┆ a   ┆ b   ┆ c   │
│ ---  ┆ --- ┆ --- ┆ --- │
│ str  ┆ str ┆ str ┆ str │
╞══════╪═════╪═════╪═════╡
│ id   ┆ a   ┆ b   ┆ c   │
│ col1 ┆ 1   ┆ 3   ┆ 2   │
│ col2 ┆ 3   ┆ 4   ┆ 6   │
└──────┴─────┴─────┴─────┘

You can add slice(1) maybe, but can we add a param like first_column_as_new_header so that everything will be taken care of:

┌──────┬─────┬─────┬─────┐
│ id   ┆ a   ┆ b   ┆ c   │
│ ---  ┆ --- ┆ --- ┆ --- │
│ str  ┆ str ┆ str ┆ str │
╞══════╪═════╪═════╪═════╡
│ col1 ┆ 1   ┆ 3   ┆ 2   │
│ col2 ┆ 3   ┆ 4   ┆ 6   │
└──────┴─────┴─────┴─────┘

This could also potentially useful for transpose a describe table

df.describe().transpose()
shape: (4, 7)
┌──────────┬────────────┬───────────────────┬────────────────────┬──────────┬──────────┬──────────┐
│ column_0 ┆ column_1   ┆ column_2          ┆ column_3           ┆ column_4 ┆ column_5 ┆ column_6 │
│ ---      ┆ ---        ┆ ---               ┆ ---                ┆ ---      ┆ ---      ┆ ---      │
│ str      ┆ str        ┆ str               ┆ str                ┆ str      ┆ str      ┆ str      │
╞══════════╪════════════╪═══════════════════╪════════════════════╪══════════╪══════════╪══════════╡
│ count    ┆ null_count ┆ mean              ┆ std                ┆ min      ┆ max      ┆ median   │
│ 3        ┆ 0          ┆ null              ┆ null               ┆ a        ┆ c        ┆ null     │
│ 3.0      ┆ 0.0        ┆ 2.0               ┆ 1.0                ┆ 1.0      ┆ 3.0      ┆ 2.0      │
│ 3.0      ┆ 0.0        ┆ 4.333333333333333 ┆ 1.5275252316519465 ┆ 3.0      ┆ 6.0      ┆ 4.0      │
└──────────┴────────────┴───────────────────┴────────────────────┴──────────┴──────────┴──────────┘

[cmdlineluser]

Not that it addresses your particular issue but .glimpse() may be of interest if you're not already aware of it.

>>> df.glimpse()
Rows: 3
Columns: 3
$ id   <str> a, b, c
$ col1 <i64> 1, 3, 2
$ col2 <i64> 3, 4, 6

>>> df.describe().glimpse()
Rows: 7
Columns: 4
$ describe <str> count, null_count, mean, std, min, max, median
$ id       <str> 3, 0, None, None, a, c, None
$ col1     <f64> 3.0, 0.0, 2.0, 1.0, 1.0, 3.0, 2.0
$ col2     <f64> 3.0, 0.0, 4.333333333333333, 1.5275252316519465, 3.0, 6.0, 4.0

[stevenlis]

@cmdlineluser Thanks for mentioning it. It would be nice if it can be read in a table.

[stevenlis]

Thanks! @magarick
This can be done with:

import polars as pl

pl.DataFrame(
    {"id": ['a', 'b', 'c'],
     "col1": [1, 3, 2],
     "col2": [3, 4, 6]}
).transpose(
    include_header=True, header_name='id', column_names='id'
)