'SVC' object has no attribute 'best_params_'

Question

def ParamSelection(X, Y, nfolds): 
    Cs = [0.0001, 0.001, 0.01, 0.1, 1, 10, 100]
    degrees = [1, 2, 3, 4, 5]
    param_grid = {'C': Cs, 'degree' : degrees}
    grid_search = GridSearchCV(svm.SVC(kernel = 'poly'), param_grid, cv = nfolds)
    grid_search = svm.SVC(gamma='scale')
    grid_search.fit(X, Y)
    grid_search.best_params_
    return grid_search.best_params_
ParamSelection(trainX, trainY, 10)

AttributeError: 'SVC' object has no attribute 'best_params_'

I'm getting the above error. What should I do? Can you help?

Answer 1

You re-assigned GridSearch to svm, so your GridSearch object is not the sklearn object. Just delete this line:

grid_search = svm.SVC(gamma='scale')

Then it should run fine. You'll get a deprecation warning, so set the gamma parameter when instantiating SVC():

grid_search = GridSearchCV(svm.SVC(kernel = 'poly', gamma='scale'), param_grid, cv = nfolds)