# summarize history for accuracy and loss
modelHistory(history)
y_pred = model.predict([x_test_rev], batch_size=1024, verbose=1)
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Testing accuracy
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
acc = sum([
np.argmax(y_test_rev[i]) == np.argmax(y_pred[i])
for i in range(x_test_rev.shape[0])
]) / x_test_rev.shape[0]
print("Model accuracy:", acc)
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Plot confusion matrix
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
y_pred_copy = np.copy(y_pred)
for row in range(y_pred_copy.shape[0]):
y_pred_copy[row] = (y_pred_copy[row] == np.amax(
y_pred_copy[row])).astype(int)
np.set_printoptions(precision=2)
plot_cf(y_test_rev.argmax(axis=1),
y_pred_copy.argmax(axis=1),
"LSTM",
normalize=True)
plt.show()
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Classification report (precision, recall, f1, micro¯o avg, weighted avg)
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
print('Classification Report')
print(classification_report(y_test_rev.argmax(axis=1), y_pred.argmax(axis=1)))
#hyperparameters tuning
#from sklearn.model_selection import GridSearchCV
#def svc_param_selection(X, y, nfolds):
# Cs = {'C': [0.001, 0.10, 0.1, 10, 25, 50]}
# grid_search = GridSearchCV(LinearSVC(), Cs, cv=nfolds)
# grid_search.fit(X, y)
# grid_search.best_params_
# return grid_search.best_params_
#params = svc_param_selection(X_train_tfidf, y_train, 5)
#best params is C = 0.1
SVM_linear.fit(X_train_tfidf, y_train)
prediction_SVM_linear = SVM_linear.predict(X_test_tfidf)
print("Model accuracy (review predict overall rating):",
accuracy_score(y_test, prediction_SVM_linear))
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Plot confusion matrix
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
plot_cf(y_test, prediction_SVM_linear, "SVM", normalize=True)
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Classification report (precision, recall, f1, micro¯o avg, weighted avg)
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
report = classification_report(y_test, prediction_SVM_linear)
print(report)
"""""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """
Save SVM model to disk
""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """""" """"""
filename = 'svm_model_rev.sav'
pickle.dump(SVM_linear, open(filename, 'wb'))