[Python] Python Polars unable to convert f64 column to str and aggregate to list

Working with a dataframe of the type:

df = pl.DataFrame({
'order': [38681.0, 38692.0, 38680.0, 38693.0],
'shipto': ["471433", "239269", "471433","239269"],
'value': [10,20,30,None]
})


Need to Groupby the 'Shipto' column, sum the 'values' and aggregate the 'order' to a list. Have tried a few things but can't get it work.

The base script:

df = (df
.with_columns([
pl.col('order').cast(pl.Utf8),
pl.col(pl.Float64).fill_null(0)
])
.groupby('shipto')
.agg([
pl.col('order').apply(lambda x: str(x)).alias('order_list'),
pl.sum('value')
])
)


Returns:

shipto order_list value
str str i64
471433 shape: (2,) 40
Series: '' [f64]
[
...
239269 shape: (2,) 20
Series: '' [f64]
[
...

What I am hoping to get in the 'order_list' column is either ([38681.0,38680.0],[38692.0,38693.0]) or (['38681.0','38680.0'],['38692.0','38693.0'])

I am guessing the 'order' column needs to be converted from f64 values to string (Utf8), but can't get that that to work.

Variations to the line 'pl.col('order').cast(pl.Utf8), #.cast(pl.Float64)' that I have tried so far:

pl.col('order').cast(pl.Float64).cast(pl.Utf8),

pl.col('order').cast(pl.Int64).cast(pl.Utf8),

pl.col('order').map(lambda x: str(x)),

pl.col('order').apply(lambda x: str(int(x))),

pl
.when(pl.col('order').is_null())
.then(pl.lit(''))
.otherwise(pl.col('order').cast(pl.Float64).cast(pl.Utf8)).alias('order'),


Sure there is some basic error, but continuing to struggle with this and any help would be greatly appreciated.

Continue reading...