We will now run GridSearch for our selected parameters. Here, we are choosing to include bigrams and trigrams while running GridSearch over the C parameter of LogisticRegression.
Our intention here is to automate as much as possible. Instead of trying varying values in C during our RandomizedSearch, we are trading off human learning time (a few hours) with compute time (a few extra minutes). This mindset saves us both time and effort.
from sklearn.model_selection import GridSearchCVparam_grid = dict(clf__C=[85, 100, 125, 150])grid_search = GridSearchCV(lr_clf, param_grid=param_grid, scoring='accuracy', n_jobs=-1, cv=3)grid_search.fit(X_train, y_train)grid_search.best_estimator_.steps
In the preceding lines of code, we have ran ...
