What is the meaning of kernel in GridSearchCV?

Question

In this code below, what is the meaning of the key kernel in the variable parameters?

from sklearn.model_selection import GridSearchCV
from sklearn.svm import SVC

parameters = {'kernel': ('linear', 'rbf'), 'C': (1, 4, 8, 16, 32)}

svc = SVC()
clf = GridSearchCV(svc, parameters, cv=5)
clf.fit(train_x_vectors, train_y)

Answer 1

The purpose of grid search is to tune the hyperparameters of a given modelling algorithm. In particular, the GridSearchCV object needs to know which hyperparameters to vary, and what values to vary them over. That's what the parameters dictionary is for. All of the keys in parameters must match hyperparameters for the particular algorithm.

In this case, kernel and C are hyperparameters for the support vector classifier (SVC) controlling the type of kernel to use, and the regularization parameter respectively. The GridSearchCV machinery is simply passing them on to the model.

Check out the SVC docs and the GridSearchCV docs