In the following dataframe I would like to multiply var_3 and var_4 by negative 1. I can do so using the following method but I am wondering if it can be done by collecting them in a list (imagining that there may be many more than 4 columns in the dataframe)
df = pl.DataFrame({"var_1": ["a", "a", "b"],
"var_2": ["c", "d", "e"],
"var_3": [1, 2, 3],
"var_4": [4, 5, 6]})
df.with_columns([pl.col("var_3") * -1,
pl.col("var_4") * -1])
Which returns the desired dataframe
shape: (3, 4)
┌───────┬───────┬───────┬───────┐
│ var_1 ┆ var_2 ┆ var_3 ┆ var_4 │
│ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ str ┆ i64 ┆ i64 │
╞═══════╪═══════╪═══════╪═══════╡
│ a ┆ c ┆ -1 ┆ -4 │
│ a ┆ d ┆ -2 ┆ -5 │
│ b ┆ e ┆ -3 ┆ -6 │
└───────┴───────┴───────┴───────┘
My try at it goes like this, though it is not applying the multiplication:
var_list = ["var_3", "var_4"]
pl_cols_var_list = [pl.col(k) for k in var_list]
df.with_columns(pl_cols_var_list * -1)
You were close. You can provide your list of variable names (as strings) directly to the polars.col expression:
var_list = ["var_3", "var_4"]
df.with_columns(pl.col(var_list) * -1)
shape: (3, 4)
┌───────┬───────┬───────┬───────┐
│ var_1 ┆ var_2 ┆ var_3 ┆ var_4 │
│ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ str ┆ i64 ┆ i64 │
╞═══════╪═══════╪═══════╪═══════╡
│ a ┆ c ┆ -1 ┆ -4 │
│ a ┆ d ┆ -2 ┆ -5 │
│ b ┆ e ┆ -3 ┆ -6 │
└───────┴───────┴───────┴───────┘
Another tip, if you have lots of columns and want to exclude only a few, you can use the polars.exclude expression:
var_list = ["var_1", "var_2"]
df.with_columns(pl.exclude(var_list) * -1)
shape: (3, 4)
┌───────┬───────┬───────┬───────┐
│ var_1 ┆ var_2 ┆ var_3 ┆ var_4 │
│ --- ┆ --- ┆ --- ┆ --- │
│ str ┆ str ┆ i64 ┆ i64 │
╞═══════╪═══════╪═══════╪═══════╡
│ a ┆ c ┆ -1 ┆ -4 │
│ a ┆ d ┆ -2 ┆ -5 │
│ b ┆ e ┆ -3 ┆ -6 │
└───────┴───────┴───────┴───────┘