Question

文档中的sample_weight作为RandomForestQuantileRegressor.fit（X，y）的可用参数，位于：https://scikit-garden.github.io/api/#skgardenquantile_1，但是在尝试执行此操作时会引发错误。它适用于DecisionTreeQuantileRegressor，但不适用于RandomForestQuantileRegressor。这是故意的吗？

示例：

import numpy as np
import skgarden 

# we create 20 points
np.random.seed(0)
X = np.r_[np.random.randn(10, 2) + [1, 1], np.random.randn(10, 2)]
y = [1] * 10 + [-1] * 10
sample_weight_last_ten = abs(np.random.randn(len(X)))
sample_weight_constant = np.ones(len(X))
# and bigger weights to some outliers
sample_weight_last_ten[15:] *= 5
sample_weight_last_ten[9] *= 15

# fit the model WITH WEIGHTS
clf_weights = skgarden.DecisionTreeQuantileRegressor(random_state=0)
clf_weights.fit(X, y, sample_weight=sample_weight_last_ten)

可以正确训练，但是在尝试时：

import numpy as np
import skgarden 

# we create 20 points
np.random.seed(0)
X = np.r_[np.random.randn(10, 2) + [1, 1], np.random.randn(10, 2)]
y = [1] * 10 + [-1] * 10
sample_weight_last_ten = abs(np.random.randn(len(X)))
sample_weight_constant = np.ones(len(X))
# and bigger weights to some outliers
sample_weight_last_ten[15:] *= 5
sample_weight_last_ten[9] *= 15

# fit the model WITH WEIGHTS
clf_weights = skgarden.quantile.RandomForestQuantileRegressor(random_state=0)
clf_weights.fit(X, y, sample_weight=sample_weight_last_ten)

此错误弹出：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-375-dfda3c67445e> in <module>
     15 # fit the model WITH WEIGHTS
     16 clf_weights = skgarden.quantile.RandomForestQuantileRegressor(random_state=0)
---> 17 clf_weights.fit(X, y, sample_weight=sample_weight_last_ten)

TypeError: fit() got an unexpected keyword argument 'sample_weight'

训练时我可以为每个样本增加权重的另一种方式是

在skgarden.quantile.RandomForestQuantileRegressor.fit（）中无法识别出sample_weight

0 个答案: