def executeProcessing(self, text, trainingFilePath, modelPath, vectorPath,
prediction_file_object):
self.log_writer.log(prediction_file_object, 'Prediction begins now')
try:
with open(vectorPath, 'rb') as f:
vectorizer = pickle.load(f)
with open(modelPath, 'rb') as f:
loaded_model = pickle.load(f)
self.log_writer.log(
prediction_file_object,
'Prediction model and vectorizer instance loaded successfully')
except Exception as e:
self.log_writer.log(
prediction_file_object,
'Prediction unsuccessful due to model loading')
raise e
text = data_preprocessing_predict(text, self.stop_words_path,
prediction_file_object)
x = vectorizer.transform(text).toarray()
prediction = loaded_model.predict(x)
self.log_writer.log(prediction_file_object, 'Predictions generated')
return prediction[0]
def executePreocessing(self, text, jsonFilePath,modelPath,vectorPath):
df_pred = pd.DataFrame([text],columns=['text'])
df_pred['text'] = data_preprocessing_predict(df_pred['text'], self.stop_words_path)
df_pred['text'] = [" ".join(value) for value in df_pred['text'].values]
with open(vectorPath,'rb') as f:
vectorizer = pickle.load(f)
with open(modelPath, 'rb') as f:
model = pickle.load(f)
pred_vector_ = vectorizer.transform(df_pred['text'])
prediction = model.predict(pred_vector_)
predictedProbability = model.predict_proba(pred_vector_)
if list(predictedProbability.flatten())[0] == list(predictedProbability.flatten())[1]:
return "UNKNOWN"
elif list(predictedProbability.flatten())[np.argmax(predictedProbability)] > .3:
return prediction
else:
return "UNKNOWN"