def start_naivebayes_classification(self, array, mode, ngrambow, minborder, maxborder, nr, tuplebows):
""" Start Naive Bayes classification learning. Return results (resultscores_tuple, N.A., N.A.)"""
nbObject = Start_NB(array, mode, self.tweet_class, True, tuplebows, ngrambow, self.CROSS_VALIDATION)
results = nbObject.start_naivebayes_evaluation(mode, minborder, maxborder, nr)
return results
def get_activity_tweets(self, inputfile, mode, ngrambow, nr, loadclassifier):
""" Extract new activity tweets from file"""
# Create classifier on trainingdata of class
(array, tuplebows) = self.setup_input_classification(mode, ngrambow, 0, 0, nr)
if "svm" in mode:
svmObject = Start_SVM(array, mode, self.tweet_class, True, tuplebows, self.CROSS_VALIDATION)
if "nb" in mode:
nbObject = Start_NB(array, mode, self.tweet_class, True, tuplebows, ngrambow, self.CROSS_VALIDATION)
# Get tweets of new data
new_tweets = {}
index = 0
newdata = csv.reader(open(inputfile, "rU"), delimiter=self.DELIMITER)
for i, row in enumerate(newdata):
if row[5] == "" and row[0].isdigit():
new_tweets[index] = row[3]
index += 1
# Preprocess new data
inputfile_filename = inputfile.split(".")[0]
self.preprocess_tweets(mode, new_tweets, inputfile_filename)
array = self.get_preprocessed_array(mode)
# Classify newdata
if "svm" in mode:
prediction = svmObject.start_classification(mode, array, loadclassifier, 0.001, 10)
if "nb" in mode:
prediction = nbObject.start_classification(mode, array, False, loadclassifier)
# Print to file
self.count_classes(prediction.tolist())
classification_filename = inputfile_filename + "_class.csv"
helpers.write_classification_to_tweetfile(prediction, 0, 5, inputfile, classification_filename)
def analysis_classification(self, mode, ngrambow, nr, loadclassifier):
""" Analyse classification of training & testdata"""
DELIMITER = "\t"
# Get tweets
all_tweets = {}
index = 0
data = csv.reader(open(self.TRAININGFILE, "rU"), delimiter=DELIMITER)
for i, row in enumerate(data):
if i == 0:
pass
else:
all_tweets[index] = row[3]
index += 1
# Create classifier on trainingdata of class
(array, tuplebows) = self.setup_input_classification(mode, ngrambow, 0, 0, nr)
if "svm" in mode:
svmObject = Start_SVM(array, mode, self.tweet_class, True, tuplebows, self.CROSS_VALIDATION)
if "nb" in mode:
nbObject = Start_NB(array, mode, self.tweet_class, True, tuplebows, ngrambow, self.CROSS_VALIDATION)
print "preprocess new data"
# Preprocess new dataata
training_filename = self.TRAININGFILE.split(".")[0]
self.preprocess_tweets(mode, self.tweets, training_filename)
array = self.get_preprocessed_array(mode)
# Classify tweets
if "svm" in mode:
prediction = svmObject.start_classification(mode, array, loadclassifier, 0.001, 10)
if "nb" in mode:
prediction = nbObject.start_classification(mode, array, False, loadclassifier)
self.count_classes(prediction.tolist())
classification_filename = training_filename + "_class.csv"
helpers.write_classification_to_tweetfile(prediction, 1, 7, self.TRAININGFILE, classification_filename)