def calculate_features(info_path, date, user):
"""
calculate features about sleep
:param whole_path:
:param date:
:return:
"""
day = date
sleep_df = bs.read_xlsx(info_path, '睡眠信息')
res = pd.DataFrame(columns=features)
temp = [0, 0, 0, 0, 0, 0, 0, 0]
for index in sleep_df.index:
raw = sleep_df.loc[index].values
if raw[1] == user and raw[7] == day:
temp[0] = get_time(raw[3])
temp[2] = get_time(raw[2])
temp[1] = temp[2] - temp[0]
temp[3] = temp[0] / temp[2]
temp[4] = 1 - temp[3]
temp[7] = (24 * 60 - temp[2]) / (24 * 60)
temp[5] = int(raw[4][:2])
temp[6] = int(raw[5][:2])
res.loc[len(res)] = temp
res.to_csv(bs.get_save_whole_name('sleep.csv'))
def create_csv():
"""
all in one
:return:
"""
temp = []
for file in files:
df = pd.read_csv(bs.get_save_whole_name(file + '.csv'))
if 'Unnamed: 0' in df.columns.values:
del df['Unnamed: 0']
_dict = {}
for x in df.columns.values:
_dict[x] = file + '_' + x
df.rename(columns=_dict, inplace=True)
temp.append(df)
temp = pd.concat(temp, axis=1)
temp.to_csv(bs.get_save_whole_name('all_inone.csv'))
def calculaue_features(whole_path, info_path, date, users):
'''
main function
:return:
'''
home_work_place = get_home_work_place(info_path, users)
df = pd.read_csv(whole_path, header=None)
location_dict = calculate_location_features(home_work_place, df.copy(),
date)
move_dict = calculate_move_features(home_work_place, df.copy(), date)
time_dict = calculate_time_features(home_work_place, df.copy(), date)
df = modify_list_Todataframe(location_dict, move_dict, time_dict)
df.to_csv(bs.get_save_whole_name('location.csv'))
def calculate_features(whole_path, date):
"""
calculate features about call information
:param whole_path:
:param date:
:return:
"""
day = date
res = pd.DataFrame(columns=features)
_feature_dict = {} # 存放我们想要特征的字典
out_num = {}
answer_num = {}
reject_num = {}
all_num = {}
for feature in features:
_feature_dict[feature] = 0 # 开始先全部赋值为0
df = pd.read_csv(whole_path, header=None)
for index in df.index: # 遍历该用户的csv
# 得到csv的一行
raw = df.loc[index].values.tolist()
if str(raw[1]).split(' ')[0] == day:
if raw[0] == 'OUTGOTING_CALL_TYPE':
_feature_dict['out_duration'] += int(str(raw[2]).split(';')[2])
_feature_dict['all_duration'] += int(str(raw[2]).split(';')[2])
out_num[str(raw[2]).split(';')[1]] = 0
all_num[str(raw[2]).split(';')[1]] = 0
if raw[0] == 'INCOMING_CALL_TYPE':
if str(raw[2]).find('answer') >= 0:
_feature_dict['answer_duration'] += int(
str(raw[2]).split(';')[2])
_feature_dict['all_duration'] += int(
str(raw[2]).split(';')[2])
answer_num[str(raw[2]).split(';')[1]] = 0
all_num[str(raw[2]).split(';')[1]] = 0
elif str(raw[2]).find('reject') >= 0:
_feature_dict['reject_duration'] += int(
str(raw[2]).split(';')[2])
_feature_dict['all_duration'] += int(
str(raw[2]).split(';')[2])
reject_num[str(raw[2]).split(';')[1]] = 0
all_num[str(raw[2]).split(';')[1]] = 0
_feature_dict['out_people_number'] = len(out_num.keys())
_feature_dict['answer_people_number'] = len(answer_num.keys())
_feature_dict['reject_people_number'] = len(reject_num.keys())
_feature_dict['all_people_number'] = len(all_num.keys())
res.loc[len(res)] = list(_feature_dict.values())
res.to_csv(bs.get_save_whole_name('call.csv'))
def predict():
"""
predict vas
:param all_inone_path:
:return:
"""
clf, scale = load_model()
df = pd.read_csv(bs.get_save_whole_name('all_inone.csv'))
if 'Unnamed: 0' in df.columns.values:
del df['Unnamed: 0']
_list = get_features_list()
df = df[_list]
x = df.loc[0].values
x = scale.transform([x])
vas = clf.predict(x)
return vas[0]
def calculate_features(whole_path, date):
"""
calculate features about screen
:param whole_path:
:param date:
:return:
"""
day = date
res = pd.DataFrame(columns=features)
_feature_dict = {} # 存放我们想要特征的字典
for feature in features:
_feature_dict[feature] = 0 # 开始先全部赋值为0
df = pd.read_csv(whole_path, header=None)
for index in df.index: # 遍历该用户的csv
# 得到csv的一行
raw = df.loc[index].values.tolist()
if str(raw[1]).split(' ')[0] == day:
if raw[0] == 'SCREEN_ACTIVE_TYPE_2':
_feature_dict['duration'] += int(raw[2])
_feature_dict['count'] += 1
_time = bs.get_whole_time_from_str(raw[1])
if _time.hour in [0, 1, 2]:
_feature_dict['0-3'] += int(raw[2])
elif _time.hour in [3, 4, 5]:
_feature_dict['3-6'] += int(raw[2])
elif _time.hour in [6, 7, 8]:
_feature_dict['6-9'] += int(raw[2])
elif _time.hour in [9, 10, 11]:
_feature_dict['9-12'] += int(raw[2])
elif _time.hour in [12, 13, 14]:
_feature_dict['12-15'] += int(raw[2])
elif _time.hour in [15, 16, 17]:
_feature_dict['15-18'] += int(raw[2])
elif _time.hour in [18, 19, 20]:
_feature_dict['18-21'] += int(raw[2])
elif _time.hour in [21, 22, 23]:
_feature_dict['21-24'] += int(raw[2])
res.loc[len(res)] = list(_feature_dict.values())
res.to_csv(bs.get_save_whole_name('screen.csv'))
def calculate_features(whole_path, date):
"""
calculate features about app
:return:
"""
day = date
features = []
for x in big_features:
for y in small_features:
features.append(x + '_' + y)
res = pd.DataFrame(columns=features)
_feature_dict = {} # 存放我们想要特征的字典
for feature in features:
_feature_dict[feature] = 0 # 开始先全部赋值为0
df = pd.read_csv(whole_path, header=None)
for index in df.index: # 遍历该用户的csv
# 得到csv的一行
raw = df.loc[index].values.tolist()
if str(raw[1]).split(' ')[0] == day:
if raw[0] == 'OPEN_ACT_2':
for big_feature in big_features:
ll = get_class(big_feature)
IsClass = False
for l in ll:
if str(raw[2]).find(l) >= 0:
IsClass = True
if IsClass:
_feature_dict[big_feature + '_' + 'duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' + 'count'] += 1
_time = bs.get_whole_time_from_str(raw[1])
if _time.hour in [0, 1, 2]:
_feature_dict[big_feature + '_' +
'0-3_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' + '0-3_count'] += 1
elif _time.hour in [3, 4, 5]:
_feature_dict[big_feature + '_' +
'3-6_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' + '3-6_count'] += 1
elif _time.hour in [6, 7, 8]:
_feature_dict[big_feature + '_' +
'6-9_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' + '6-9_count'] += 1
elif _time.hour in [9, 10, 11]:
_feature_dict[big_feature + '_' +
'9-12_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' +
'9-12_count'] += 1
elif _time.hour in [12, 13, 14]:
_feature_dict[big_feature + '_' +
'12-15_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' +
'12-15_count'] += 1
elif _time.hour in [15, 16, 17]:
_feature_dict[big_feature + '_' +
'15-18_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' +
'15-18_count'] += 1
elif _time.hour in [18, 19, 20]:
_feature_dict[big_feature + '_' +
'18-21_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' +
'18-21_count'] += 1
elif _time.hour in [21, 22, 23]:
_feature_dict[big_feature + '_' +
'21-24_duration'] += int(
str(raw[2]).split(';')[-1])
_feature_dict[big_feature + '_' +
'21-24_count'] += 1
res.loc[len(res)] = list(_feature_dict.values())
res.to_csv(bs.get_save_whole_name('app.csv'))