dic[week][gender][hour] += 1
dic[week][gender]["all"] += 1
dic["all"][gender][hour] += 1
dic["all"][gender]["all"] += 1
if birthYear != "":
generation = cm.generation_distinction(birthYear)
dic["all"][generation][hour] += 1
dic["all"][generation]["all"] += 1
dic[week][generation][hour] += 1
dic[week][generation]["all"] += 1
if vertical != "other":
dic[week][vertical][hour] += 1
dic[week][vertical]["all"] += 1
dic["all"][vertical][hour] += 1
dic["all"][vertical]["all"] += 1
print "======================================="
print "============= all ============="
print "======================================="
cm.print_dict_content(dic["all"])
print "======================================="
print "========== weekday ============"
print "======================================="
cm.print_dict_content(dic["week"])
print "======================================="
print "========== holiday ============"
print "======================================="
cm.print_dict_content(dic["holi"])
browseId = featureArray[0]
device = featureArray[8]
vertical = cm.typical_vertical(featureArray[10])
if vertical == "other" and featureArray[10] not in other_verical_array:
other_verical_array.append(featureArray[10])
birthYear = featureArray[2]
gender = featureArray[1]
# vertical = "None"
allDict["all"]["all"] += 1
allDict["all"][vertical] += 1
if birthYear != "" and gender != "" and device in ["smartphone", "tablet", "pc", "tv", "featurephone"]:
generation = cm.generation_distinction(birthYear)
generationDict[generation]["all"] += 1
generationDict[generation][vertical] += 1
outputs["generation"][generation].write(line)
genderDict[gender]["all"] += 1
genderDict[gender][vertical] += 1
outputs["gender"][gender].write(line)
deviceDict[device]["all"] += 1
deviceDict[device][vertical] += 1
outputs["device"][device].write(line)
cm.print_dict_content(allDict)
cm.print_dict_content(deviceDict)
cm.print_dict_content(generationDict)
cm.print_dict_content(genderDict)
print "===========other============"
for x in other_device:
print x
if birthYear != "" and gender != "" and device in ["smartphone", "tablet", "pc", "tv", "featurephone"]:
if is_holiday:
generation = cm.generation_distinction(birthYear)
holidayDict["generationDict"][generation]["all"] += 1
holidayDict["generationDict"][generation][vertical] += 1
holidayDict["deviceDict"][device]["all"] += 1
holidayDict["deviceDict"][device][vertical] += 1
holidayDict["genderDict"][gender]["all"] += 1
holidayDict["genderDict"][gender][vertical] += 1
else:
generation = cm.generation_distinction(birthYear)
weekdayDict["generationDict"][generation]["all"] += 1
weekdayDict["generationDict"][generation][vertical] += 1
weekdayDict["deviceDict"][device]["all"] += 1
weekdayDict["deviceDict"][device][vertical] += 1
weekdayDict["genderDict"][gender]["all"] += 1
weekdayDict["genderDict"][gender][vertical] += 1
cm.print_dict_content(weekdayDict["allDict"])
cm.print_dict_content(weekdayDict["deviceDict"])
cm.print_dict_content(weekdayDict["generationDict"])
cm.print_dict_content(weekdayDict["genderDict"])
cm.print_dict_content(holidayDict["allDict"])
cm.print_dict_content(holidayDict["deviceDict"])
cm.print_dict_content(holidayDict["generationDict"])
cm.print_dict_content(holidayDict["genderDict"])
print "===========other============"
for x in other_device:
print x
dic = {
"all": {"all": 0},
"1": defaultdict(lambda: 0),
"2": defaultdict(lambda: 0),
"3": defaultdict(lambda: 0),
"4": defaultdict(lambda: 0),
"5": defaultdict(lambda: 0),
}
for fileNum in xrange(2, 9):
# for line in open('../data/'+str(date)+'_all', 'r'):
for line in open("../data/small_data.txt", "r"):
featureArray = line.split("\t")
if featureArray[10] == "":
continue
gender = featureArray[1]
birthYear = featureArray[2]
query = featureArray[7].replace(" ", " ")
searchTime = featureArray[5].split(",")[0]
device = featureArray[8]
vertical = cm.typical_vertical(featureArray[10])
length = len(query.split(" "))
if length > 5:
length = 5
dic["all"]["all"] += 1
dic[str(length)][vertical] += 1
dic[str(length)]["all"] += 1
cm.print_dict_content(dic)
