advanced multi conditional list comprehension

Question

I am new to Python and would like to assign a value based on mathematical operations, eg "right" if >, "left" if <, "equal" if ==, within a list comprehension.

I have tried the below, but it throws an error. Can multiple conditions be specified in a single list comprehension in this way, where each "elif" generates a different output, or will I need to use a loop?

Fully reproduceable example:

from sklearn.datasets import load_iris

bunch = load_iris(as_frame=True)
df = bunch.data.reset_index().rename(columns={"index": "id"}).merge(bunch.target.reset_index().rename(columns={"index": "id"})).drop(["id"], axis=1)

# question is in last row, "skew"
datasummary_dct = {
   "50%": [df[col].median().round(2) if any(t in str(df[col].dtype) for t in ("float", "int", "time")) else " " for col in df.columns],
   "mean": [df[col].mean().round(2) if any(t in str(df[col].dtype) for t in ("float", "int", "time")) else " " for col in df.columns],
   "skew": ["left" if df[col].median() > df[col].mean() else "right" if df[col].median() < df[col].mean() else "equal" if df[col].median()==df[col].mean() if any(t in str(df[col].dtype) for t in ("float", "int", "time")) else " " for col in df.columns],
}

again I am still fairly new to programming; apologies if I do not immediately understand the solution. any guidance is appreciated!

Answer 1

Instead of the complex nested if, you can use np.select for this which is much more readable:

datasummary_dct = {
    "skew": [
        np.select(
            [df[col].median() > df[col].mean(), df[col].median() < df[col].mean()],
            ["right", "left"],
            "equal",
        )
        if any(t in str(df[col].dtype) for t in ("float", "int", "time"))
        else " "
        for col in df.columns
    ],
}

print(pd.DataFrame(datasummary_dct))

Output:

    skew
0   left
1   left
2  right
3  right
4  equal