How to use FunctionTransformer with a custom function?

Question

I want to use FunctionTransformer to perform calculations between columns. For instance, I want to substract two columns and add the the new column to the dataset. So I have the function:

def diff(x, y):
    return x - y

my initial dataset is:

X = pd.DataFrame({"product":["a","b","c","d"], "ndp":[100,200,150,120], "discount":[5,10,15,30]})

  product  ndp  discount
0       a  100         5
1       b  200        10
2       c  150        15
3       d  120        30

and I need a new column price = ndp - discount, so I run:

from sklearn.preprocessing import FunctionTransformer

transf = FunctionTransformer(diff, kw_args={'x': 'ndp', 'y':"discount"})
func_transf.transform(X)

but I get an error:

TypeError: diff() got multiple values for argument 'x'

How can I pass the arguments to the function diff and how to specify the name of the new column?

Arne · Accepted Answer · 2023-02-24 16:27:46Z

1

Note that FunctionTransformer, like most scikit-learn classes, is designed to work with NumPy arrays, rather than whole dataframes. So your function should also take in and return a NumPy array.

For example:

import numpy as np
import pandas as pd
from sklearn.preprocessing import FunctionTransformer

def diff(X: np.array) -> np.array:
    d = (X[:, 0] - X[:, 1]).reshape(-1, 1)
    return np.concatenate((X, d), axis=1)

X = pd.DataFrame({"product": ["a", "b", "c", "d"], 
                  "ndp": [100, 200, 150, 120], 
                  "discount": [5, 10, 15, 30]})

transf = FunctionTransformer(diff) 
X_new = transf.transform(X[['ndp', 'discount']].values)

print(X_new)

[[100   5  95]
 [200  10 190]
 [150  15 135]
 [120  30  90]]

answered Feb 24, 2023 at 16:27

Arne

10.6k2 gold badges22 silver badges31 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

How to use FunctionTransformer with a custom function?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related