testX_mult = []
testY_mult = []
for p in range(len(cons_pred)):
if cons_pred[p] == 1:
testX_mult.append(testX[p])
testY_mult.append(multtestY[p])
print("Obtained multi-class training data")
testX_mult = array(testX_mult)
testY_mult = array(testY_mult)
print("Creating and Training Multi-Classification Models Now ...")
history, g_mult_prediction = multi_models.gru_multi(
trainX_mult, trainY_mult, valX_mult, valY_mult, testX_mult, testY_mult,
e)
history, l_mult_prediction = multi_models.lstm_multi(
trainX_mult, trainY_mult, valX_mult, valY_mult, testX_mult, testY_mult,
e)
history, b_mult_prediction = multi_models.blstm_multi(
trainX_mult, trainY_mult, valX_mult, valY_mult, testX_mult, testY_mult,
e)
"""f_mes = f1_score(testY_mult,g_mult_prediction,average='weighted')
#gfile.write("F-Score: "+str(f_mes)+'\n')
afile.write("GRU F-Score: "+str(f_mes)+'\n')
f_mes = f1_score(testY_mult,l_mult_prediction,average='weighted')
print(f_mes)
#lfile.write("F-Score: "+str(f_mes)+'\n')
afile.write("LSTM F-Score: "+str(f_mes)+'\n')
testX.append(padded_docs[ts])
testY.append(labels[ts])
# VALIDATION DATA
valsize = len(trainX) - math.floor(0.2 * len(trainX))
valX = array(trainX[valsize:len(trainX)])
valY = array(trainY[valsize:len(trainX)])
trainX = array(trainX[:valsize])
trainY = array(trainY[:valsize])
testX = array(testX)
testY = array(testY)
##GRU MODEL
print("Creating and Training GRU Model ...")
history, pred = multi_models.gru_multi(trainX, trainY, valX, valY, testX,
testY, e)
#del pred
f_mes = f1_score(testY, pred, average='weighted')
a_mes = accuracy_score(testY, pred)
p_mes = precision_score(testY, pred, average='weighted')
r_mes = recall_score(testY, pred, average='weighted')
rcount = 0
racism_count = 0
scount = 0
sexism_count = 0
nbcount = 0
non_bully_count = 0
for x in range(len(testX)):
if testY[x] == 1: