如何修复sklearn中的折叠？

Question

我在几个预测任务中应用 CV，并且希望始终对我的每个参数集使用相同的折叠 - 如果可能的话，也可以在不同的 python 脚本中使用，因为性能确实取决于折叠。 我正在使用 sklearns KFold：

kf = KFold(n_splits=folds, shuffle=False, random_state=1986)

并通过

构建我的折叠

for idx_split, (train_index, test_index) in enumerate(kf.split(X, Y)):
    X_train, X_test = X[train_index], X[test_index]
    y_train, y_test = Y[train_index], Y[test_index]

然后像这样循环遍历它们

for idx_alpha, alpha in enumerate([0, 0.2, 0.4, 0.6, 0.8, 1]):
    # [...]
    for idx_split, (train_index, test_index) in enumerate(kf.split(X, Y)):
        X_train, X_test = X[train_index], X[test_index]
        y_train, y_test = Y[train_index], Y[test_index]**

虽然我选择了一个 random_state 并设置了一个 numpy 种子，但折叠并不总是相等。我可以做些什么来实现这一点，并可能通过几个 python 脚本共享我的折叠？

Answer 1

你似乎在重塑 GridSearchCV ;-)

试试这个方法：

from sklearn.model_selection import GridSearchCV

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25)

param_grid = dict(model__alpha=[0, 0.2, 0.4, 0.6, 0.8, 1])

model = Lasso()  # put here algorithm, that you want to use

folds = 3
# alternatively you can prepare folds yourself
#folds = KFold(n_splits=folds, shuffle=False, random_state=1986)
grid_search = GridSearchCV(model, param_grid=param_grid, cv=folds, n_jobs=-1, verbose=2)
grid_search.fit(X_train, y_train)

y_pred = grid_search.best_estimator_.predict(X_test)