如何使用sklearn的TransformedTargetRegressor与自定义的数据转换器？

Question

我想使用sklearn.compose.TransformedTargetRegressor，就像这个回答所示。然而，转换器是自定义的，我遇到了错误。

在这个简单的例子中，目标值应该乘以十，然后在预测时再除以十。(在我的实际应用中，目标值必须从非数值格式转换为数值格式。)

import numpy as np
import sklearn
from sklearn.compose import TransformedTargetRegressor
from sklearn.linear_model import LinearRegression

class MyTransform(sklearn.base.TransformerMixin):
    def fit(self, *_, **__):
        return self

    def transform(self, X):
        return np.array(X)*10

    def inverse_transform(self, X):
        return np.array(X)/10


def MyLinearRegression():
    return TransformedTargetRegressor(
        regressor=LinearRegression(),
        transformer=MyTransform()
    )


if __name__ == '__main__':
    model = MyLinearRegression()
    model.fit(X=[[1], [2], [3]], y=[1, 2, 3]) # raises TypeError

这引起了：

Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2019.3.3\plugins\python-ce\helpers\pydev\_pydev_bundle\pydev_umd.py", line 197, in runfile
    pydev_imports.execfile(filename, global_vars, local_vars)  # execute the script
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2019.3.3\plugins\python-ce\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "C:/Users/me/.PyCharmCE2019.3/config/scratches/scratch.py", line 26, in <module>
    model.fit(X=[[1], [2], [3]], y=[1, 2, 3]) # raises TypeError
  File "C:\Users\me\.virtualenvs\project--3333Ox_\lib\site-packages\sklearn\compose\_target.py", line 185, in fit
    self._fit_transformer(y_2d)
  File "C:\Users\me\.virtualenvs\project--3333Ox_\lib\site-packages\sklearn\compose\_target.py", line 127, in _fit_transformer
    self.transformer_ = clone(self.transformer)
  File "C:\Users\me\.virtualenvs\project--3333Ox_\lib\site-packages\sklearn\base.py", line 64, in clone
    raise TypeError("Cannot clone object '%s' (type %s): "
TypeError: Cannot clone object '<__main__.MyTransform object at 0x000001653B2FA9A0>' (type <class '__main__.MyTransform'>): it does not seem to be a scikit-learn estimator as it does not implement a 'get_params' methods.

- actual_panda

Answer 1

你只需要继承自sklearn.base.BaseEstimator以及transformermixin : )。类型错误提示如下：

看起来它不是一个scikit-learn的评估器

所以你只需要把它变成一个评估器即可 :D。以下代码应该可以工作。

import numpy as np
import sklearn
from sklearn.compose import TransformedTargetRegressor
from sklearn.linear_model import LinearRegression

class MyTransform(sklearn.base.BaseEstimator, sklearn.base.TransformerMixin):
    def fit(self, *_, **__):
        return self

    def transform(self, X):
        return np.array(X)*10

    def inverse_transform(self, X):
        return np.array(X)/10


def MyLinearRegression():
    return TransformedTargetRegressor(
        regressor=LinearRegression(),
        transformer=MyTransform()
    )



model = MyLinearRegression()
model.fit(X=[[1], [2], [3]], y=[1, 2, 3])

- nbertagnolli