Scikit learn 参数具有弹性值误差的交叉Val分数
我试图使用Scikit learn 参数具有弹性值误差的交叉Val分数,scikit-learn,cross-validation,valueerror,Scikit Learn,Cross Validation,Valueerror,我试图使用cross\u val\u score()函数和嵌套交叉验证在python中实现一个简单的弹性网络回归,但它不允许我传递参数。它一直在为我的l1_比率声明一个无效参数的ValueError,我不明白为什么给定它在0和1之间 ValueError: Invalid parameter l1_ratio for estimator Pipeline(steps=[('preprocessor', ColumnTransformer(remainder='p
cross\u val\u score()
函数和嵌套交叉验证在python中实现一个简单的弹性网络回归,但它不允许我传递参数。它一直在为我的l1_比率声明一个无效参数的ValueError,我不明白为什么给定它在0和1之间
ValueError: Invalid parameter l1_ratio for estimator Pipeline(steps=[('preprocessor',
ColumnTransformer(remainder='passthrough',
transformers=[('cat', OneHotEncoder(), [0]),
('num', StandardScaler(),
slice(1, 37, None))])),
('model', ElasticNet(random_state=42))]).
Check the list of available parameters with `estimator.get_params().keys()`.
我的代码:
cv_outer=KFold(10, shuffle=True)
cv_inner=KFold(10, shuffle=True)
models_params = {
'en': (LREN(random_state=42), # Elastic Net
{'l1_ratio': [0,0.25,0.5,0.75,1]
,'alpha':[1e-2,1e-1,1,1e1]})
# My first column is Categorical, the other 36 are numerical
preprocessor = ColumnTransformer(
transformers=[
('cat', OneHotEncoder(), [0])
,('num', StandardScaler(), slice(1,37))
]
,remainder = 'passthrough')
# Store Results
average_scores = dict()
for name, (model, params) in models_params.items():
mymodel = Pipeline(steps = [('preprocessor', preprocessor),
('model', model)
])
# this object is a regressor that also happens to choose
# its hyperparameters automatically using `inner_cv`
optimize_hparams = GridSearchCV(
estimator = mymodel, param_grid=params, n_jobs = -1,
cv=cv_inner, scoring='neg_mean_absolute_error')
# estimate generalization error on the outer-fold splits of the data
outer_folds_scores = cross_val_score(
optimize_hparams,
X, y, cv=cv_outer, scoring='neg_mean_absolute_error')
您可以尝试定义网格,例如:
models_params = {
{'model__l1_ratio': [0,0.25,0.5,0.75,1],
'model__alpha':[1e-2,1e-1,1,1e1]}
}
啊,是的,我记得其他车型需要在其他地方这样做。小挫折。是的,你需要这样做,当你有很多步骤在你的管道