Update values ​​in DF based on column names

Question content

I have the next pandas dataframe:

x_1 x_2 x_3 x_4 col_to_replace cor_factor
    
    1   2   3   4   x_2            1
    3   3   5   1   x_1            6
    2   2   0   0   x_3            0
...

I want to update the name column saved in col_to_replace with the value in cor_factor and save the result in the corresponding column as well as the car_factor column. Some (ugly) solutions might be:

for i in len(df.shape[0]):
    df[df['col_to_replace']].iloc[i] = df[df['col_to_replace']].iloc[i] - df['cor_factor'].iloc[i]                                                                          
    df['cor_factor'].iloc[i] = df['cor_factor'].iloc[i] -  df[df['col_to_replace']].iloc[i]

This method is definitely not time-saving. I'm looking for a faster solution.

The output of df should be like this:

x_1 x_2 x_3 x_4 col_to_replace cor_factor
    
    1   1   3   4   x_2            -1
    -3  3   5   1   x_1            3
    2   2   0   0   x_3            0
...

Correct answer

Use pivot to correct the x_ value and index lookup Correct the last column. Since the value changes, make sure to copy before modifying:

# perform indexing lookup
# save the value for later
idx, cols = pd.factorize(df['col_to_replace'])
corr = df.reindex(cols, axis=1).to_numpy()[np.arange(len(df)), idx]

# pivot and subtract the factor
# ensure original order of the columns
cols = df.columns.intersection(cols, sort=false)
df[cols] = df[cols].sub(df.pivot(columns='col_to_replace',
                                 values='cor_factor'),
                        fill_value=0).convert_dtypes()
# correct with the saved "corr"
df['cor_factor'] -= corr

Output:

x_1  x_2  x_3  x_4 col_to_replace  cor_factor
0    1    1    3    4            x_2          -1
1   -3    3    5    1            x_1           3
2    2    2    0    0            x_3           0

The above is the detailed content of Update values ​​in DF based on column names. For more information, please follow other related articles on the PHP Chinese website!