Skip to content

Instantly share code, notes, and snippets.

@hakanserce

hakanserce/HyperParameterOptimization.py

Created May 12, 2017 17:04
Show Gist options
  • Save hakanserce/9cac8d291af679331da265839b3b7400 to your computer and use it in GitHub Desktop.
Save hakanserce/9cac8d291af679331da265839b3b7400 to your computer and use it in GitHub Desktop.
Download ZIP
Using KFold cross correlation for hyper-parameter optimization...
Raw
HyperParameterOptimization.py
# This gist uses RBF Kernel based Ridge Regression: https://gist.github.com/hakanserce/fdd571132ef44a6a8f7ddd2eb41aba84
kf = KFold(y_train.size, n_folds=5)
X_all = np.hstack((X_train, X_test))
y_all = np.hstack((y_train, y_test))
def get_cross_validated_mses(rbf_lambda, d2):
for train, test in kf:
X_train_cv, X_test_cv, y_train_cv, y_test_cv = X_all[train], X_all[test], y_all[train], y_all[test]
# train your model
model_cv = BasisFunctionRidgeRegression(RBFKernelTransform(X_train_cv, rbf_lambda), d2)
model_cv.train(X_train_cv, y_train_cv)
# return MSE
yield model_cv.mse(X_test_cv, y_test_cv)
def get_cross_validated_mse(rbf_lambda, d2):
mses = np.fromiter(get_cross_validated_mses(rbf_lambda,d2), dtype=float)
return np.mean(mses)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment