使用熊猫组合列集

Question

I have the following data frame structure:

    SC0 Shape   S1  S2  S3  C1  C2  C3  D1  D2  D3
2   1   Circle  NaN NaN NaN 1   1   1   NaN NaN NaN
3   13  Square  2   1   2   NaN NaN NaN NaN NaN NaN
4   13  Diamond NaN NaN NaN NaN NaN NaN 2   1   2
5   16  Diamond NaN NaN NaN NaN NaN NaN 2   2   2
6   16  Square  2   2   2   NaN NaN NaN NaN NaN NaN

How can I combine S1,S2,S3 with C1,C2,C3,D1,D2,D3 so S1,C1 and D1 are on the same column, S2,C2 and D2...(all the way to S16,C16 and D16)?

When Shape = Circle the populated columns are C1-C16, when Shape = Square its S1-S16 and for Shape = Diamond its D1-D16.

I don't mind creating a new set of columns or copy two of them to to an existing set, as long as I have all the #1 scores in the same column, #2 same column etc.

Thank you!

Answer 1

IIUC you have an equal amount of columns for each category, and you want to compress this into numeric columns which are shape agnostic. If so this will work:

dfs = []
for var in ['S', 'D', 'C']:
        # filter  columns with a regex
        res = df[df.iloc[:, 2:].filter(regex= var + '\d{1,2}').columns].dropna()
        # rename coumns with just numbers to enable concatenation
        res.columns = range(3)
        dfs.append(res)

df = pd.concat([df.iloc[:, :2], pd.concat(dfs)], 1)
print(df)

Output:

   SC0  Shape        0      1       2
2   1   Circle      1.0     1.0     1.0
3   13  Square      2.0     1.0     2.0
4   13  Diamond     2.0     1.0     2.0
5   16  Diamond     2.0     2.0     2.0
6   16  Square      2.0     2.0     2.0

Answer 2

Try:

n=3
cols_prefixes=["C", "S", "D"]
for i in range(n):
    cols=[f"{l}{i+1}" for l in cols_prefixes]
    df[f"res{i+1}"]=df[cols].bfill(axis=1).iloc[:,0]
    df=df.drop(columns=cols)

Outputs:

   SC0    Shape  res1  res2  res3
2    1   Circle   1.0   1.0   1.0
3   13   Square   2.0   1.0   2.0
4   13  Diamond   2.0   1.0   2.0
5   16  Diamond   2.0   2.0   2.0
6   16   Square   2.0   2.0   2.0