I have a pandas dataframe for example

I want to pick the minimum absolute value: table2

For now I am using df.abs().idxmin(axis="columns") and I get: table3

I would like to ask how can I convert table 3 to table 2?

CodePudding user response：

Use np.argmin (numpy counterpart of DataFrame.idxmin). Since you want to extract the original values, it's more convenient to access those values at the numpy level.

I added an extra row to your MRE for demonstration:

cols = np.argmin(df.abs().to_numpy(), axis=1) # [0, 2]
rows = range(len(cols))                       # [0, 1]
df['Result'] = df.to_numpy()[rows, cols]

#    Col1  Col2  Col3  Result
# 0    -1     2    -3      -1
# 1    10    -5     4       4

CodePudding user response：

You can use df.abs().min(axis=1)

>>> import pandas as pd
>>>
>>> data = {'col1': [1], 'col2': [2], 'col3': [-3]}
>>> df = pd.DataFrame(data)
>>> df
   col1  col2  col3
0     1     2    -3
>>> df['Result'] = df.abs().min(axis=1)
>>> df
   col1  col2  col3  Result
0     1     2    -3       1

CodePudding user response：

Try:

df = pd.read_clipboard()
print(df)
#    Col1  Col2  Col3
# 0    -1     2    -3

df['Result'] = df.abs().idxmin(axis=1)
print(df)
#    Col1  Col2  Col3 Result
# 0    -1     2    -3   Col1

df['Result'] = df.loc[:, df['Result']]
print(df)
#    Col1  Col2  Col3  Result
# 0    -1     2    -3      -1

Output:

   Col1  Col2  Col3  Result
0    -1     2    -3      -1