Define a function that extract a float and do operation

I have a column with the following type of values:

I have to define a function that convert column Diameter in float and manage that kind of exception. In particular:

• when is 42×54 then make the operation: sqrt(42^2+54^2)
• when is Steel then return a NaN value

>Solution :

You can use str.extract with a regex to get the numbers, then square them with pow, sum the columns, and get the square root with numpy.sqrt:

import numpy as np

df['Diameter2'] = np.sqrt(df['Diameter']
                          .str.extract('(\d+)(?:\s*x\s*(\d+))?')
                          .astype(float).pow(2)
                          .sum(axis=1, min_count=1)
                          )

output:

  Diameter  Diameter2
0       44  44.000000
1  42 x 54  68.410526
2    Steel        NaN

regex demo

(\d+)              # capture a number
(?:\s*x\s*(\d+))?  # capture a number (optionally) if preceded by x with optional spaces