Edward Edward - 1 month ago 48
Python Question

_transform() takes 2 positional arguments but 3 were given

I try to build a pipeline with variable transformation
And i do as below

import numpy as np
import pandas as pd
import sklearn
from sklearn import linear_model
from sklearn.base import BaseEstimator, TransformerMixin
from sklearn.pipeline import Pipeline


Dataframe

df = pd.DataFrame({'y': [4,5,6], 'a':[3,2,3], 'b' : [2,3,4]})


I try to get a new variable for predict

class Complex():
def __init__(self, X1, X2):
self.a = X1
self.b = X2
def transform(self, X1, X2):
age = pd.DataFrame(self.a - self.b)
return age
def fit_transform(self, X1, X2):
self.fit( X1, X2)
return self.transform(X1, X2)

def fit(self, X1, X2):
return self


Then i make a pipeline

X = df[['a', 'b']]
y = df['y']
regressor = linear_model.SGDRegressor()
pipeline = Pipeline([
('transform', Complex(X['a'], X['b'])) ,
('model_fitting', regressor)
])
pipeline.fit(X, y)


and i get error

pred = pipeline.predict(X)
pred
TypeError Traceback (most recent call last)
<ipython-input-555-7a07ccb0c38a> in <module>()
----> 1 pred = pipeline.predict(X)
2 pred

C:\Program Files\Anaconda3\lib\site-packages\sklearn\utils\metaestimators.py in <lambda>(*args, **kwargs)
52
53 # lambda, but not partial, allows help() to work with update_wrapper
---> 54 out = lambda *args, **kwargs: self.fn(obj, *args, **kwargs)
55 # update the docstring of the returned function
56 update_wrapper(out, self.fn)

C:\Program Files\Anaconda3\lib\site-packages\sklearn\pipeline.py in predict(self, X)
324 for name, transform in self.steps[:-1]:
325 if transform is not None:
--> 326 Xt = transform.transform(Xt)
327 return self.steps[-1][-1].predict(Xt)
328

TypeError: transform() missing 1 required positional argument: 'X2'


what i do wrong? I see the mistake is in class Complex(). How to fix it?

Answer

So the problem is that transform expects an argument of array of shape [n_samples, n_features]

See the Examples section in the documentation of sklearn.pipeline.Pipeline, it uses sklearn.feature_selection.SelectKBest as a transform, and you can see its source that it expects X to be an array instead of separate variables like X1 and X2.

In short, your code can be fixed like this:


import pandas as pd
import sklearn
from sklearn import linear_model
from sklearn.pipeline import Pipeline

df = pd.DataFrame({'y': [4,5,6], 'a':[3,2,3], 'b' : [2,3,4]})

class Complex():
    def transform(self, Xt):
        return pd.DataFrame(Xt['a'] - Xt['b'])

    def fit_transform(self, X1, X2):
        return self.transform(X1)

X = df[['a', 'b']]
y = df['y']
regressor = linear_model.SGDRegressor()
pipeline = Pipeline([
        ('transform', Complex()) ,
        ('model_fitting', regressor)
    ])
pipeline.fit(X, y)

pred = pipeline.predict(X)
print(pred)