def linear_regression(X_train, X_test, Y_train, Y_test, plot=False):
# Create linear regression object
# The training data should be column vectors
X_train, X_test = np.array(X_train).reshape((len(X_train), 1)), np.array(X_test).reshape((len(X_test), 1))
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(X_train, Y_train)
predict = regr.predict(X_test)
return regression_evaluate(Y_test, predict)
def linear_regression(X_train, X_test, Y_train, Y_test, plot=False):
# Create linear regression object
# The training data should be column vectors
X_train, X_test = np.array(X_train).reshape(
(len(X_train), 1)), np.array(X_test).reshape((len(X_test), 1))
regr = linear_model.LinearRegression()
# Train the model using the training sets
regr.fit(X_train, Y_train)
predict = regr.predict(X_test)
return regression_evaluate(Y_test, predict)
print('The true values is: %s' % true_values[i])
return arithmetic, geometric
if __name__ == '__main__':
########################################### Hyper-parameters ###########################################
target = 'arousal' # values: "valence", "arousal"
categorical = 'all' # values: 'all', "book", "car", "laptop", "hotel", "news", "political"
########################################################################################################
# texts, valence, arousal = read_mix_data(categorical)
from load_data import load_CVAT_3
# texts, valence, arousal = load_CVAT_3('./resources/corpus 2009 sigma 1.5.csv','./resources/tokenized_texts.p', categorical=categorical)
texts, valence, arousal = load_CVAT_3('./resources/CVAT (utf-8).csv','./resources/tokenized_texts_(newest3.31).p', categorical=categorical)
lexicon = load_CVAW()
d = dict()
if target == 'valence':
ind = 1
true_values = valence
print('Valence prediction...')
elif target == 'arousal':
ind = 2
true_values = arousal
print('Arousal preddiction...')
for l in lexicon:
d[l[0]] = l[ind]
arithmetic, geometric = va_prediction(texts, d, true_values)
print('Prediction result (arithmetic average):')
regression_evaluate(true_values, arithmetic)
print('Prediction result (geometric average):')
regression_evaluate(true_values, geometric)
from mix_data import read_mix_data
texts, valence, arousal = read_mix_data(categorical)
if option == 'V':
Y = valence
elif option == 'A':
Y = arousal
else:
raise Exception('Wrong parameters!')
lexicon = load_CVAW(extended=using_extended_lexicon)
d = dict()
ind = 1 if option == 'V' else 2
for l in lexicon:
d[l[0]] = l[ind]
predicted_ratings = mean_ratings(texts, d, mean_method, Y)
print(predicted_ratings)
print(Y)
out = regression_evaluate(Y, predicted_ratings)
draw_scatter(Y, predicted_ratings, 'True Values', 'Predicted Values', title='Scatter')
out2 = cv(predicted_ratings, Y)
Dims = 'Valence' if option == 'V' else 'Arousal'
Mean_Method = 'Geometric' if mean_method == 'tf_geo' else 'Arithmetic'
print('|%s|%s|False|%.3f|%.3f|%.3f|' % (Dims, Mean_Method, out[0], out[1], out[2]))
print('|%s|%s|True|%.3f|%.3f|%.3f|' % (Dims, Mean_Method, out2[0], out2[1], out2[2]))
Y = valence
elif option == 'A':
Y = arousal
else:
raise Exception('Wrong parameters!')
lexicon = load_CVAW(extended=using_extended_lexicon)
d = dict()
ind = 1 if option == 'V' else 2
for l in lexicon:
d[l[0]] = l[ind]
predicted_ratings = mean_ratings(texts, d, mean_method, Y)
print(predicted_ratings)
print(Y)
out = regression_evaluate(Y, predicted_ratings)
draw_scatter(Y,
predicted_ratings,
'True Values',
'Predicted Values',
title='Scatter')
out2 = cv(predicted_ratings, Y)
Dims = 'Valence' if option == 'V' else 'Arousal'
Mean_Method = 'Geometric' if mean_method == 'tf_geo' else 'Arithmetic'
print('|%s|%s|False|%.3f|%.3f|%.3f|' %
(Dims, Mean_Method, out[0], out[1], out[2]))
print('|%s|%s|True|%.3f|%.3f|%.3f|' %
(Dims, Mean_Method, out2[0], out2[1], out2[2]))
if __name__ == '__main__':
########################################### Hyper-parameters ###########################################
target = 'arousal' # values: "valence", "arousal"
categorical = 'all' # values: 'all', "book", "car", "laptop", "hotel", "news", "political"
########################################################################################################
# texts, valence, arousal = read_mix_data(categorical)
from load_data import load_CVAT_3
# texts, valence, arousal = load_CVAT_3('./resources/corpus 2009 sigma 1.5.csv','./resources/tokenized_texts.p', categorical=categorical)
texts, valence, arousal = load_CVAT_3(
'./resources/CVAT (utf-8).csv',
'./resources/tokenized_texts_(newest3.31).p',
categorical=categorical)
lexicon = load_CVAW()
d = dict()
if target == 'valence':
ind = 1
true_values = valence
print('Valence prediction...')
elif target == 'arousal':
ind = 2
true_values = arousal
print('Arousal preddiction...')
for l in lexicon:
d[l[0]] = l[ind]
arithmetic, geometric = va_prediction(texts, d, true_values)
print('Prediction result (arithmetic average):')
regression_evaluate(true_values, arithmetic)
print('Prediction result (geometric average):')
regression_evaluate(true_values, geometric)