def tenFoldSplitCSV(src_csv, dst_path): with open(src_csv, 'rb') as csvfile: if not os.path.exists(dst_path): os.makedirs(dst_path) length = 0 lines = [] reader = csv.reader(csvfile, delimiter=',') for row in reader: lines.append(row) length+=1 kf = KFold(length, n_folds = 10) dir_iter = 1 for train, test in kf: foldcsv = "Train\n" for train_index in train: foldcsv = foldcsv + csvLineMaker(lines[train_index]) + "\n" foldcsv = foldcsv + "Test\n" for test_index in test: foldcsv = foldcsv + csvLineMaker(lines[train_index]) + "\n" saveFile(dst_path + "fold" + str(dir_iter) + ".csv", foldcsv) dir_iter+=1
def extractDataFilesAndFeatures(initialFolderpath,finalFolderPath, resultFolderPath):
csvTextualFeature = csvLineMaker(getTextualHeaderFeatures())+'\n'
csvTemporalFeature = csvLineMaker(getTemporalHeaderFeatures())+'\n'
csvClassFeature = csvLineMaker(getClassHeaderFeatures())+'\n'
fileNames = getFilesFromFolder(initialFolderpath)
for fileName in fileNames:
content, textFeatures, temporalFeatures = extractTextAndFeatures(initialFolderpath+'/'+fileName)
info = fileName.split('.')
userID = str(info[0])
gender = str(info[1])
age = str(info[2])
ageClass = str(extractAgeClass(float(age)))
job = str(info[3])
place = str(info[4])
saveFile(finalFolderPath+'/'+userID+'.txt', content)
csvTextLine= csvLineMaker(textFeatures)
classFeatures = []
classFeatures.append(userID)
classFeatures.append(age)
classFeatures.append(gender)
classFeatures.append(job)
classFeatures.append(place)
classFeatures.append(ageClass)
csvClassLine = csvLineMaker(classFeatures)
csvTemporalLine= csvLineMaker(temporalFeatures)
csvClassFeature = csvClassFeature+csvClassLine+'\n'
csvTextualFeature = csvTextualFeature+csvTextLine+'\n'
csvTemporalFeature = csvTemporalFeature+csvTemporalLine+'\n'
saveFile(resultFolderPath+'/textualFeature.csv', csvTextualFeature)
saveFile(resultFolderPath+'/temporalFeature.csv', csvTemporalFeature)
saveFile(resultFolderPath+'/classFeature.csv', csvClassFeature)