CODEWITHSUNDEEP
python-polars
Horseman
Horseman

Reputation: 417

How to rename column names with first row in polars?

I have a polars dataframe and I want to set the first row as header. I thought about renaming the column name one by one with the value of the first row of the correspondent column. How can I do this in polars?

Upvotes: 0

Views: 4709

Answers (2)

Dean MacGregor
Dean MacGregor

Reputation: 18446

Another way to do this is:

df.columns=df.iter_rows().__next__()

iter_rows makes a generator so it's not going to read any more of the df than the first row. If you don't like using the internal __next__ then you could do

df.columns=df.head(1).rows()[0]

or

for row in df.iter_rows():
    df.columns=row
    break

or

rows=df.iter_rows()
df.columns=next(rows)

Upvotes: 0

jqurious
jqurious

Reputation: 21279

[UPDATE]: @kirara0048's suggestion of .to_dicts() is a much simpler approach.

>>> df.head(1).to_dicts().pop()
{'column_0': 'one', 'column_1': 'two', 'column_2': 'three'}

Which can be passed directly to .rename()

df.rename(df.head(1).to_dicts().pop())

Perhaps there is a more direct method but you could take the first row and .transpose().to_series()

>>> df = pl.DataFrame([["one", "four"], ["two", "five"], ["three", "six"]])
>>> df.head(1).transpose().to_series()
shape: (3,)
Series: 'column_0' [str]
[
    "one"
    "two"
    "three"
]

This can be used to create a dict of old: new and passed to .rename()

>>> df.rename(dict(zip(df.columns, df.head(1).transpose().to_series())))
shape: (2, 3)
┌──────┬──────┬───────┐
│ one  | two  | three │
│ ---  | ---  | ---   │
│ str  | str  | str   │
╞══════╪══════╪═══════╡
│ one  | two  | three │
├──────┼──────┼───────┤
│ four | five | six   │
└──────┴──────┴───────┘

.slice(1) can be used to "remove" the first row if desired:

>>> df.rename(dict(zip(df.columns, df.head(1).transpose().to_series()))).slice(1)
shape: (1, 3)
┌──────┬──────┬───────┐
│ one  | two  | three │
│ ---  | ---  | ---   │
│ str  | str  | str   │
╞══════╪══════╪═══════╡
│ four | five | six   │
└──────┴──────┴───────┘

You can also assign to .columns - I'm unsure if this is considered "bad style" or not.

>>> df.columns = df.head(1).transpose().to_series()
>>> df
shape: (2, 3)
┌──────┬──────┬───────┐
│ one  | two  | three │
│ ---  | ---  | ---   │
│ str  | str  | str   │
╞══════╪══════╪═══════╡
│ one  | two  | three │
├──────┼──────┼───────┤
│ four | five | six   │
└──────┴──────┴───────┘

Upvotes: 2

Related Questions