1 def get_train_data():
2 df = pd.read_csv('data/train.csv', encoding='utf_8')
3
4 # df1 = pd.read_csv('data/test.csv', encoding='utf_8')
5 # df2 = pd.read_csv('data/count.csv', encoding='utf_8')
6 # df1['casual'] = df2['casual']
7 # df1['registered'] = df2['registered']
8 # df1['count'] = df2['count']
9 # df = pd.concat([df, df1])
10
11 cur_time = pd.to_datetime(df['datetime'])
12
13 df['year'] = cur_time.map(lambda x:x.year)
14 df['month'] = cur_time.map(lambda x:x.month)
15 df['day'] = cur_time.map(lambda x:x.day)
16 df['hour'] = cur_time.map(lambda x:x.hour)
17 df['dayOfWeek'] = cur_time.map(lambda x:x.isoweekday())
18 df['segOfDay'] = df['hour'].apply(classfy)
19
20 year = pd.get_dummies(data=df.iloc[:,])
21
22 # df['spring'] = df['season'].map({1:1})
23 # df['summer'] = df['season'].map({2:1})
24 # df['autumn'] = df['season'].map({3:1})
25 # df['winter'] = df['season'].map({4:1})
26 # df = df.fillna(value={'spring':0,'summer':0,'autumn':0,'winter':0})
27 #
28 # df['weather1'] = df['weather'].map({1:1})
29 # df['weather2'] = df['weather'].map({2:1})
30 # df['weather3'] = df['weather'].map({3:1})
31 # df['weather4'] = df['weather'].map({4:1})
32 # df = df.fillna(value={'weather1':0,'weather2':0,'weather3':0,'weather4':0})
33 #
34 df = df.replace({'windspeed':0}, 12.799)
35
36 casual = df['casual']
37 registered = df['registered']
38 df = df.drop(['datetime', 'season', 'weather','casual', 'registered', 'count'], axis=1)
39
40 log_cas = casual.map(lambda x: math.log(x+1))
41 log_reg = registered.map(lambda x: math.log(x+1))
42
43 train_casual = log_cas.values
44 train_registered = log_reg.values
45 train_data = df.values
46
47 return train_casual, train_registered, train_data
