关闭

七月机器学习项目实战之特征工程6城市自行车共享系统使用状况

标签: 机器学习
1197人阅读 评论(0) 收藏 举报
分类:

特征工程小案例

Kaggle上有这样一个比赛:城市自行车共享系统使用状况。

提供的数据为2年内按小时做的自行车租赁数据,其中训练集由每个月的前19天组成,测试集由20号之后的时间组成。

本项目功能:数据清理,特征提取,标准化连续值特征,离散性数据实现one-hot编码
本项目数据及源码:https://github.com/qiu997018209/MachineLearning

#先把数据读进来
import pandas as pd
data = pd.read_csv('kaggle_bike_competition_train.csv', header = 0, error_bad_lines=False)
#看一眼数据长什么样
data.head()
datetime season holiday workingday weather temp atemp humidity windspeed casual registered count
0 2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0 3 13 16
1 2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0 8 32 40
2 2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0 5 27 32
3 2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0 3 10 13
4 2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0 0 1 1
### 把datetime域切成 日期 和 时间 两部分。
# 处理时间字段
temp = pd.DatetimeIndex(data['datetime'])
data['date'] = temp.date
data['time'] = temp.time
data.head()
datetime season holiday workingday weather temp atemp humidity windspeed casual registered count date time
0 2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0 3 13 16 2011-01-01 00:00:00
1 2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0 8 32 40 2011-01-01 01:00:00
2 2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0 5 27 32 2011-01-01 02:00:00
3 2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0 3 10 13 2011-01-01 03:00:00
4 2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0 0 1 1 2011-01-01 04:00:00
### 时间那部分,好像最细的粒度也只到小时,所以我们干脆把小时字段拿出来作为更简洁的特征。
# 设定hour这个小时字段
data['hour'] = pd.to_datetime(data.time, format="%H:%M:%S")
data['hour'] = pd.Index(data['hour']).hour
data
datetime season holiday workingday weather temp atemp humidity windspeed casual registered count date time hour
0 2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0000 3 13 16 2011-01-01 00:00:00 0
1 2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0000 8 32 40 2011-01-01 01:00:00 1
2 2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0000 5 27 32 2011-01-01 02:00:00 2
3 2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0000 3 10 13 2011-01-01 03:00:00 3
4 2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0000 0 1 1 2011-01-01 04:00:00 4
5 2011-01-01 05:00:00 1 0 0 2 9.84 12.880 75 6.0032 0 1 1 2011-01-01 05:00:00 5
6 2011-01-01 06:00:00 1 0 0 1 9.02 13.635 80 0.0000 2 0 2 2011-01-01 06:00:00 6
7 2011-01-01 07:00:00 1 0 0 1 8.20 12.880 86 0.0000 1 2 3 2011-01-01 07:00:00 7
8 2011-01-01 08:00:00 1 0 0 1 9.84 14.395 75 0.0000 1 7 8 2011-01-01 08:00:00 8
9 2011-01-01 09:00:00 1 0 0 1 13.12 17.425 76 0.0000 8 6 14 2011-01-01 09:00:00 9
10 2011-01-01 10:00:00 1 0 0 1 15.58 19.695 76 16.9979 12 24 36 2011-01-01 10:00:00 10
11 2011-01-01 11:00:00 1 0 0 1 14.76 16.665 81 19.0012 26 30 56 2011-01-01 11:00:00 11
12 2011-01-01 12:00:00 1 0 0 1 17.22 21.210 77 19.0012 29 55 84 2011-01-01 12:00:00 12
13 2011-01-01 13:00:00 1 0 0 2 18.86 22.725 72 19.9995 47 47 94 2011-01-01 13:00:00 13
14 2011-01-01 14:00:00 1 0 0 2 18.86 22.725 72 19.0012 35 71 106 2011-01-01 14:00:00 14
15 2011-01-01 15:00:00 1 0 0 2 18.04 21.970 77 19.9995 40 70 110 2011-01-01 15:00:00 15
16 2011-01-01 16:00:00 1 0 0 2 17.22 21.210 82 19.9995 41 52 93 2011-01-01 16:00:00 16
17 2011-01-01 17:00:00 1 0 0 2 18.04 21.970 82 19.0012 15 52 67 2011-01-01 17:00:00 17
18 2011-01-01 18:00:00 1 0 0 3 17.22 21.210 88 16.9979 9 26 35 2011-01-01 18:00:00 18
19 2011-01-01 19:00:00 1 0 0 3 17.22 21.210 88 16.9979 6 31 37 2011-01-01 19:00:00 19
20 2011-01-01 20:00:00 1 0 0 2 16.40 20.455 87 16.9979 11 25 36 2011-01-01 20:00:00 20
21 2011-01-01 21:00:00 1 0 0 2 16.40 20.455 87 12.9980 3 31 34 2011-01-01 21:00:00 21
22 2011-01-01 22:00:00 1 0 0 2 16.40 20.455 94 15.0013 11 17 28 2011-01-01 22:00:00 22
23 2011-01-01 23:00:00 1 0 0 2 18.86 22.725 88 19.9995 15 24 39 2011-01-01 23:00:00 23
24 2011-01-02 00:00:00 1 0 0 2 18.86 22.725 88 19.9995 4 13 17 2011-01-02 00:00:00 0
25 2011-01-02 01:00:00 1 0 0 2 18.04 21.970 94 16.9979 1 16 17 2011-01-02 01:00:00 1
26 2011-01-02 02:00:00 1 0 0 2 17.22 21.210 100 19.0012 1 8 9 2011-01-02 02:00:00 2
27 2011-01-02 03:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 4 6 2011-01-02 03:00:00 3
28 2011-01-02 04:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 1 3 2011-01-02 04:00:00 4
29 2011-01-02 06:00:00 1 0 0 3 17.22 21.210 77 19.9995 0 2 2 2011-01-02 06:00:00 6
10856 2012-12-18 18:00:00 4 0 1 1 15.58 19.695 46 22.0028 13 512 525 2012-12-18 18:00:00 18
10857 2012-12-18 19:00:00 4 0 1 1 15.58 19.695 46 26.0027 19 334 353 2012-12-18 19:00:00 19
10858 2012-12-18 20:00:00 4 0 1 1 14.76 16.665 50 16.9979 4 264 268 2012-12-18 20:00:00 20
10859 2012-12-18 21:00:00 4 0 1 1 14.76 17.425 50 15.0013 9 159 168 2012-12-18 21:00:00 21
10860 2012-12-18 22:00:00 4 0 1 1 13.94 16.665 49 0.0000 5 127 132 2012-12-18 22:00:00 22
10861 2012-12-18 23:00:00 4 0 1 1 13.94 17.425 49 6.0032 1 80 81 2012-12-18 23:00:00 23
10862 2012-12-19 00:00:00 4 0 1 1 12.30 15.910 61 0.0000 6 35 41 2012-12-19 00:00:00 0
10863 2012-12-19 01:00:00 4 0 1 1 12.30 15.910 65 6.0032 1 14 15 2012-12-19 01:00:00 1
10864 2012-12-19 02:00:00 4 0 1 1 11.48 15.150 65 6.0032 1 2 3 2012-12-19 02:00:00 2
10865 2012-12-19 03:00:00 4 0 1 1 10.66 13.635 75 8.9981 0 5 5 2012-12-19 03:00:00 3
10866 2012-12-19 04:00:00 4 0 1 1 9.84 12.120 75 8.9981 1 6 7 2012-12-19 04:00:00 4
10867 2012-12-19 05:00:00 4 0 1 1 10.66 14.395 75 6.0032 2 29 31 2012-12-19 05:00:00 5
10868 2012-12-19 06:00:00 4 0 1 1 9.84 12.880 75 6.0032 3 109 112 2012-12-19 06:00:00 6
10869 2012-12-19 07:00:00 4 0 1 1 10.66 13.635 75 8.9981 3 360 363 2012-12-19 07:00:00 7
10870 2012-12-19 08:00:00 4 0 1 1 9.84 12.880 87 7.0015 13 665 678 2012-12-19 08:00:00 8
10871 2012-12-19 09:00:00 4 0 1 1 11.48 14.395 75 7.0015 8 309 317 2012-12-19 09:00:00 9
10872 2012-12-19 10:00:00 4 0 1 1 13.12 16.665 70 7.0015 17 147 164 2012-12-19 10:00:00 10
10873 2012-12-19 11:00:00 4 0 1 1 16.40 20.455 54 15.0013 31 169 200 2012-12-19 11:00:00 11
10874 2012-12-19 12:00:00 4 0 1 1 16.40 20.455 54 19.0012 33 203 236 2012-12-19 12:00:00 12
10875 2012-12-19 13:00:00 4 0 1 1 17.22 21.210 50 12.9980 30 183 213 2012-12-19 13:00:00 13
10876 2012-12-19 14:00:00 4 0 1 1 17.22 21.210 50 12.9980 33 185 218 2012-12-19 14:00:00 14
10877 2012-12-19 15:00:00 4 0 1 1 17.22 21.210 50 19.0012 28 209 237 2012-12-19 15:00:00 15
10878 2012-12-19 16:00:00 4 0 1 1 17.22 21.210 50 23.9994 37 297 334 2012-12-19 16:00:00 16
10879 2012-12-19 17:00:00 4 0 1 1 16.40 20.455 50 26.0027 26 536 562 2012-12-19 17:00:00 17
10880 2012-12-19 18:00:00 4 0 1 1 15.58 19.695 50 23.9994 23 546 569 2012-12-19 18:00:00 18
10881 2012-12-19 19:00:00 4 0 1 1 15.58 19.695 50 26.0027 7 329 336 2012-12-19 19:00:00 19
10882 2012-12-19 20:00:00 4 0 1 1 14.76 17.425 57 15.0013 10 231 241 2012-12-19 20:00:00 20
10883 2012-12-19 21:00:00 4 0 1 1 13.94 15.910 61 15.0013 4 164 168 2012-12-19 21:00:00 21
10884 2012-12-19 22:00:00 4 0 1 1 13.94 17.425 61 6.0032 12 117 129 2012-12-19 22:00:00 22
10885 2012-12-19 23:00:00 4 0 1 1 13.12 16.665 66 8.9981 4 84 88 2012-12-19 23:00:00 23

10886 rows × 15 columns

### 仔细想想,数据只告诉我们是哪天了,按照一般逻辑,应该周末和工作日出去的人数量不同吧。我们设定一个新的字段dayofweek表示是一周中的第几天。再设定一个字段dateDays表示离第一天开始租车多久了(猜测在欧美国家,这种绿色环保的出行方式,会迅速蔓延吧)
# 我们对时间类的特征做处理,产出一个星期几的类别型变量
data['dayofweek'] = pd.DatetimeIndex(data.date).dayofweek

# 对时间类特征处理,产出一个时间长度变量
data['dateDays'] = (data.date - data.date[0]).astype('timedelta64[D]')

data
datetime season holiday workingday weather temp atemp humidity windspeed casual registered count date time hour dayofweek dateDays
0 2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0000 3 13 16 2011-01-01 00:00:00 0 5 0.0
1 2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0000 8 32 40 2011-01-01 01:00:00 1 5 0.0
2 2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0000 5 27 32 2011-01-01 02:00:00 2 5 0.0
3 2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0000 3 10 13 2011-01-01 03:00:00 3 5 0.0
4 2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0000 0 1 1 2011-01-01 04:00:00 4 5 0.0
5 2011-01-01 05:00:00 1 0 0 2 9.84 12.880 75 6.0032 0 1 1 2011-01-01 05:00:00 5 5 0.0
6 2011-01-01 06:00:00 1 0 0 1 9.02 13.635 80 0.0000 2 0 2 2011-01-01 06:00:00 6 5 0.0
7 2011-01-01 07:00:00 1 0 0 1 8.20 12.880 86 0.0000 1 2 3 2011-01-01 07:00:00 7 5 0.0
8 2011-01-01 08:00:00 1 0 0 1 9.84 14.395 75 0.0000 1 7 8 2011-01-01 08:00:00 8 5 0.0
9 2011-01-01 09:00:00 1 0 0 1 13.12 17.425 76 0.0000 8 6 14 2011-01-01 09:00:00 9 5 0.0
10 2011-01-01 10:00:00 1 0 0 1 15.58 19.695 76 16.9979 12 24 36 2011-01-01 10:00:00 10 5 0.0
11 2011-01-01 11:00:00 1 0 0 1 14.76 16.665 81 19.0012 26 30 56 2011-01-01 11:00:00 11 5 0.0
12 2011-01-01 12:00:00 1 0 0 1 17.22 21.210 77 19.0012 29 55 84 2011-01-01 12:00:00 12 5 0.0
13 2011-01-01 13:00:00 1 0 0 2 18.86 22.725 72 19.9995 47 47 94 2011-01-01 13:00:00 13 5 0.0
14 2011-01-01 14:00:00 1 0 0 2 18.86 22.725 72 19.0012 35 71 106 2011-01-01 14:00:00 14 5 0.0
15 2011-01-01 15:00:00 1 0 0 2 18.04 21.970 77 19.9995 40 70 110 2011-01-01 15:00:00 15 5 0.0
16 2011-01-01 16:00:00 1 0 0 2 17.22 21.210 82 19.9995 41 52 93 2011-01-01 16:00:00 16 5 0.0
17 2011-01-01 17:00:00 1 0 0 2 18.04 21.970 82 19.0012 15 52 67 2011-01-01 17:00:00 17 5 0.0
18 2011-01-01 18:00:00 1 0 0 3 17.22 21.210 88 16.9979 9 26 35 2011-01-01 18:00:00 18 5 0.0
19 2011-01-01 19:00:00 1 0 0 3 17.22 21.210 88 16.9979 6 31 37 2011-01-01 19:00:00 19 5 0.0
20 2011-01-01 20:00:00 1 0 0 2 16.40 20.455 87 16.9979 11 25 36 2011-01-01 20:00:00 20 5 0.0
21 2011-01-01 21:00:00 1 0 0 2 16.40 20.455 87 12.9980 3 31 34 2011-01-01 21:00:00 21 5 0.0
22 2011-01-01 22:00:00 1 0 0 2 16.40 20.455 94 15.0013 11 17 28 2011-01-01 22:00:00 22 5 0.0
23 2011-01-01 23:00:00 1 0 0 2 18.86 22.725 88 19.9995 15 24 39 2011-01-01 23:00:00 23 5 0.0
24 2011-01-02 00:00:00 1 0 0 2 18.86 22.725 88 19.9995 4 13 17 2011-01-02 00:00:00 0 6 1.0
25 2011-01-02 01:00:00 1 0 0 2 18.04 21.970 94 16.9979 1 16 17 2011-01-02 01:00:00 1 6 1.0
26 2011-01-02 02:00:00 1 0 0 2 17.22 21.210 100 19.0012 1 8 9 2011-01-02 02:00:00 2 6 1.0
27 2011-01-02 03:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 4 6 2011-01-02 03:00:00 3 6 1.0
28 2011-01-02 04:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 1 3 2011-01-02 04:00:00 4 6 1.0
29 2011-01-02 06:00:00 1 0 0 3 17.22 21.210 77 19.9995 0 2 2 2011-01-02 06:00:00 6 6 1.0
10856 2012-12-18 18:00:00 4 0 1 1 15.58 19.695 46 22.0028 13 512 525 2012-12-18 18:00:00 18 1 717.0
10857 2012-12-18 19:00:00 4 0 1 1 15.58 19.695 46 26.0027 19 334 353 2012-12-18 19:00:00 19 1 717.0
10858 2012-12-18 20:00:00 4 0 1 1 14.76 16.665 50 16.9979 4 264 268 2012-12-18 20:00:00 20 1 717.0
10859 2012-12-18 21:00:00 4 0 1 1 14.76 17.425 50 15.0013 9 159 168 2012-12-18 21:00:00 21 1 717.0
10860 2012-12-18 22:00:00 4 0 1 1 13.94 16.665 49 0.0000 5 127 132 2012-12-18 22:00:00 22 1 717.0
10861 2012-12-18 23:00:00 4 0 1 1 13.94 17.425 49 6.0032 1 80 81 2012-12-18 23:00:00 23 1 717.0
10862 2012-12-19 00:00:00 4 0 1 1 12.30 15.910 61 0.0000 6 35 41 2012-12-19 00:00:00 0 2 718.0
10863 2012-12-19 01:00:00 4 0 1 1 12.30 15.910 65 6.0032 1 14 15 2012-12-19 01:00:00 1 2 718.0
10864 2012-12-19 02:00:00 4 0 1 1 11.48 15.150 65 6.0032 1 2 3 2012-12-19 02:00:00 2 2 718.0
10865 2012-12-19 03:00:00 4 0 1 1 10.66 13.635 75 8.9981 0 5 5 2012-12-19 03:00:00 3 2 718.0
10866 2012-12-19 04:00:00 4 0 1 1 9.84 12.120 75 8.9981 1 6 7 2012-12-19 04:00:00 4 2 718.0
10867 2012-12-19 05:00:00 4 0 1 1 10.66 14.395 75 6.0032 2 29 31 2012-12-19 05:00:00 5 2 718.0
10868 2012-12-19 06:00:00 4 0 1 1 9.84 12.880 75 6.0032 3 109 112 2012-12-19 06:00:00 6 2 718.0
10869 2012-12-19 07:00:00 4 0 1 1 10.66 13.635 75 8.9981 3 360 363 2012-12-19 07:00:00 7 2 718.0
10870 2012-12-19 08:00:00 4 0 1 1 9.84 12.880 87 7.0015 13 665 678 2012-12-19 08:00:00 8 2 718.0
10871 2012-12-19 09:00:00 4 0 1 1 11.48 14.395 75 7.0015 8 309 317 2012-12-19 09:00:00 9 2 718.0
10872 2012-12-19 10:00:00 4 0 1 1 13.12 16.665 70 7.0015 17 147 164 2012-12-19 10:00:00 10 2 718.0
10873 2012-12-19 11:00:00 4 0 1 1 16.40 20.455 54 15.0013 31 169 200 2012-12-19 11:00:00 11 2 718.0
10874 2012-12-19 12:00:00 4 0 1 1 16.40 20.455 54 19.0012 33 203 236 2012-12-19 12:00:00 12 2 718.0
10875 2012-12-19 13:00:00 4 0 1 1 17.22 21.210 50 12.9980 30 183 213 2012-12-19 13:00:00 13 2 718.0
10876 2012-12-19 14:00:00 4 0 1 1 17.22 21.210 50 12.9980 33 185 218 2012-12-19 14:00:00 14 2 718.0
10877 2012-12-19 15:00:00 4 0 1 1 17.22 21.210 50 19.0012 28 209 237 2012-12-19 15:00:00 15 2 718.0
10878 2012-12-19 16:00:00 4 0 1 1 17.22 21.210 50 23.9994 37 297 334 2012-12-19 16:00:00 16 2 718.0
10879 2012-12-19 17:00:00 4 0 1 1 16.40 20.455 50 26.0027 26 536 562 2012-12-19 17:00:00 17 2 718.0
10880 2012-12-19 18:00:00 4 0 1 1 15.58 19.695 50 23.9994 23 546 569 2012-12-19 18:00:00 18 2 718.0
10881 2012-12-19 19:00:00 4 0 1 1 15.58 19.695 50 26.0027 7 329 336 2012-12-19 19:00:00 19 2 718.0
10882 2012-12-19 20:00:00 4 0 1 1 14.76 17.425 57 15.0013 10 231 241 2012-12-19 20:00:00 20 2 718.0
10883 2012-12-19 21:00:00 4 0 1 1 13.94 15.910 61 15.0013 4 164 168 2012-12-19 21:00:00 21 2 718.0
10884 2012-12-19 22:00:00 4 0 1 1 13.94 17.425 61 6.0032 12 117 129 2012-12-19 22:00:00 22 2 718.0
10885 2012-12-19 23:00:00 4 0 1 1 13.12 16.665 66 8.9981 4 84 88 2012-12-19 23:00:00 23 2 718.0

10886 rows × 17 columns

### 其实我们刚才一直都在猜测,并不知道真实的日期相关的数据分布对吧,所以我们要做一个小小的统计来看看真实的数据分布,我们统计一下一周各天的自行车租赁情况(分注册的人和没注册的人)
byday = data.groupby('dayofweek')
# 统计下没注册的用户租赁情况
byday['casual'].sum().reset_index()
dayofweek casual
0 0 46288
1 1 35365
2 2 34931
3 3 37283
4 4 47402
5 5 100782
6 6 90084
# 统计下注册的用户的租赁情况
byday['registered'].sum().reset_index()
dayofweek registered
0 0 249008
1 1 256620
2 2 257295
3 3 269118
4 4 255102
5 5 210736
6 6 195462
### 周末既然有不同,就单独拿一列出来给星期六,再单独拿一列出来给星期日
data['Saturday']=0
data.Saturday[data.dayofweek==5]=1

data['Sunday']=0
data.Sunday[data.dayofweek==6]=1

data
/opt/conda/envs/python2/lib/python2.7/site-packages/ipykernel/__main__.py:2: SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy from ipykernel import kernelapp as app /opt/conda/envs/python2/lib/python2.7/site-packages/ipykernel/__main__.py:5: SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
datetime season holiday workingday weather temp atemp humidity windspeed casual registered count date time hour dayofweek dateDays Saturday Sunday
0 2011-01-01 00:00:00 1 0 0 1 9.84 14.395 81 0.0000 3 13 16 2011-01-01 00:00:00 0 5 0.0 1 0
1 2011-01-01 01:00:00 1 0 0 1 9.02 13.635 80 0.0000 8 32 40 2011-01-01 01:00:00 1 5 0.0 1 0
2 2011-01-01 02:00:00 1 0 0 1 9.02 13.635 80 0.0000 5 27 32 2011-01-01 02:00:00 2 5 0.0 1 0
3 2011-01-01 03:00:00 1 0 0 1 9.84 14.395 75 0.0000 3 10 13 2011-01-01 03:00:00 3 5 0.0 1 0
4 2011-01-01 04:00:00 1 0 0 1 9.84 14.395 75 0.0000 0 1 1 2011-01-01 04:00:00 4 5 0.0 1 0
5 2011-01-01 05:00:00 1 0 0 2 9.84 12.880 75 6.0032 0 1 1 2011-01-01 05:00:00 5 5 0.0 1 0
6 2011-01-01 06:00:00 1 0 0 1 9.02 13.635 80 0.0000 2 0 2 2011-01-01 06:00:00 6 5 0.0 1 0
7 2011-01-01 07:00:00 1 0 0 1 8.20 12.880 86 0.0000 1 2 3 2011-01-01 07:00:00 7 5 0.0 1 0
8 2011-01-01 08:00:00 1 0 0 1 9.84 14.395 75 0.0000 1 7 8 2011-01-01 08:00:00 8 5 0.0 1 0
9 2011-01-01 09:00:00 1 0 0 1 13.12 17.425 76 0.0000 8 6 14 2011-01-01 09:00:00 9 5 0.0 1 0
10 2011-01-01 10:00:00 1 0 0 1 15.58 19.695 76 16.9979 12 24 36 2011-01-01 10:00:00 10 5 0.0 1 0
11 2011-01-01 11:00:00 1 0 0 1 14.76 16.665 81 19.0012 26 30 56 2011-01-01 11:00:00 11 5 0.0 1 0
12 2011-01-01 12:00:00 1 0 0 1 17.22 21.210 77 19.0012 29 55 84 2011-01-01 12:00:00 12 5 0.0 1 0
13 2011-01-01 13:00:00 1 0 0 2 18.86 22.725 72 19.9995 47 47 94 2011-01-01 13:00:00 13 5 0.0 1 0
14 2011-01-01 14:00:00 1 0 0 2 18.86 22.725 72 19.0012 35 71 106 2011-01-01 14:00:00 14 5 0.0 1 0
15 2011-01-01 15:00:00 1 0 0 2 18.04 21.970 77 19.9995 40 70 110 2011-01-01 15:00:00 15 5 0.0 1 0
16 2011-01-01 16:00:00 1 0 0 2 17.22 21.210 82 19.9995 41 52 93 2011-01-01 16:00:00 16 5 0.0 1 0
17 2011-01-01 17:00:00 1 0 0 2 18.04 21.970 82 19.0012 15 52 67 2011-01-01 17:00:00 17 5 0.0 1 0
18 2011-01-01 18:00:00 1 0 0 3 17.22 21.210 88 16.9979 9 26 35 2011-01-01 18:00:00 18 5 0.0 1 0
19 2011-01-01 19:00:00 1 0 0 3 17.22 21.210 88 16.9979 6 31 37 2011-01-01 19:00:00 19 5 0.0 1 0
20 2011-01-01 20:00:00 1 0 0 2 16.40 20.455 87 16.9979 11 25 36 2011-01-01 20:00:00 20 5 0.0 1 0
21 2011-01-01 21:00:00 1 0 0 2 16.40 20.455 87 12.9980 3 31 34 2011-01-01 21:00:00 21 5 0.0 1 0
22 2011-01-01 22:00:00 1 0 0 2 16.40 20.455 94 15.0013 11 17 28 2011-01-01 22:00:00 22 5 0.0 1 0
23 2011-01-01 23:00:00 1 0 0 2 18.86 22.725 88 19.9995 15 24 39 2011-01-01 23:00:00 23 5 0.0 1 0
24 2011-01-02 00:00:00 1 0 0 2 18.86 22.725 88 19.9995 4 13 17 2011-01-02 00:00:00 0 6 1.0 0 1
25 2011-01-02 01:00:00 1 0 0 2 18.04 21.970 94 16.9979 1 16 17 2011-01-02 01:00:00 1 6 1.0 0 1
26 2011-01-02 02:00:00 1 0 0 2 17.22 21.210 100 19.0012 1 8 9 2011-01-02 02:00:00 2 6 1.0 0 1
27 2011-01-02 03:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 4 6 2011-01-02 03:00:00 3 6 1.0 0 1
28 2011-01-02 04:00:00 1 0 0 2 18.86 22.725 94 12.9980 2 1 3 2011-01-02 04:00:00 4 6 1.0 0 1
29 2011-01-02 06:00:00 1 0 0 3 17.22 21.210 77 19.9995 0 2 2 2011-01-02 06:00:00 6 6 1.0 0 1
10856 2012-12-18 18:00:00 4 0 1 1 15.58 19.695 46 22.0028 13 512 525 2012-12-18 18:00:00 18 1 717.0 0 0
10857 2012-12-18 19:00:00 4 0 1 1 15.58 19.695 46 26.0027 19 334 353 2012-12-18 19:00:00 19 1 717.0 0 0
10858 2012-12-18 20:00:00 4 0 1 1 14.76 16.665 50 16.9979 4 264 268 2012-12-18 20:00:00 20 1 717.0 0 0
10859 2012-12-18 21:00:00 4 0 1 1 14.76 17.425 50 15.0013 9 159 168 2012-12-18 21:00:00 21 1 717.0 0 0
10860 2012-12-18 22:00:00 4 0 1 1 13.94 16.665 49 0.0000 5 127 132 2012-12-18 22:00:00 22 1 717.0 0 0
10861 2012-12-18 23:00:00 4 0 1 1 13.94 17.425 49 6.0032 1 80 81 2012-12-18 23:00:00 23 1 717.0 0 0
10862 2012-12-19 00:00:00 4 0 1 1 12.30 15.910 61 0.0000 6 35 41 2012-12-19 00:00:00 0 2 718.0 0 0
10863 2012-12-19 01:00:00 4 0 1 1 12.30 15.910 65 6.0032 1 14 15 2012-12-19 01:00:00 1 2 718.0 0 0
10864 2012-12-19 02:00:00 4 0 1 1 11.48 15.150 65 6.0032 1 2 3 2012-12-19 02:00:00 2 2 718.0 0 0
10865 2012-12-19 03:00:00 4 0 1 1 10.66 13.635 75 8.9981 0 5 5 2012-12-19 03:00:00 3 2 718.0 0 0
10866 2012-12-19 04:00:00 4 0 1 1 9.84 12.120 75 8.9981 1 6 7 2012-12-19 04:00:00 4 2 718.0 0 0
10867 2012-12-19 05:00:00 4 0 1 1 10.66 14.395 75 6.0032 2 29 31 2012-12-19 05:00:00 5 2 718.0 0 0
10868 2012-12-19 06:00:00 4 0 1 1 9.84 12.880 75 6.0032 3 109 112 2012-12-19 06:00:00 6 2 718.0 0 0
10869 2012-12-19 07:00:00 4 0 1 1 10.66 13.635 75 8.9981 3 360 363 2012-12-19 07:00:00 7 2 718.0 0 0
10870 2012-12-19 08:00:00 4 0 1 1 9.84 12.880 87 7.0015 13 665 678 2012-12-19 08:00:00 8 2 718.0 0 0
10871 2012-12-19 09:00:00 4 0 1 1 11.48 14.395 75 7.0015 8 309 317 2012-12-19 09:00:00 9 2 718.0 0 0
10872 2012-12-19 10:00:00 4 0 1 1 13.12 16.665 70 7.0015 17 147 164 2012-12-19 10:00:00 10 2 718.0 0 0
10873 2012-12-19 11:00:00 4 0 1 1 16.40 20.455 54 15.0013 31 169 200 2012-12-19 11:00:00 11 2 718.0 0 0
10874 2012-12-19 12:00:00 4 0 1 1 16.40 20.455 54 19.0012 33 203 236 2012-12-19 12:00:00 12 2 718.0 0 0
10875 2012-12-19 13:00:00 4 0 1 1 17.22 21.210 50 12.9980 30 183 213 2012-12-19 13:00:00 13 2 718.0 0 0
10876 2012-12-19 14:00:00 4 0 1 1 17.22 21.210 50 12.9980 33 185 218 2012-12-19 14:00:00 14 2 718.0 0 0
10877 2012-12-19 15:00:00 4 0 1 1 17.22 21.210 50 19.0012 28 209 237 2012-12-19 15:00:00 15 2 718.0 0 0
10878 2012-12-19 16:00:00 4 0 1 1 17.22 21.210 50 23.9994 37 297 334 2012-12-19 16:00:00 16 2 718.0 0 0
10879 2012-12-19 17:00:00 4 0 1 1 16.40 20.455 50 26.0027 26 536 562 2012-12-19 17:00:00 17 2 718.0 0 0
10880 2012-12-19 18:00:00 4 0 1 1 15.58 19.695 50 23.9994 23 546 569 2012-12-19 18:00:00 18 2 718.0 0 0
10881 2012-12-19 19:00:00 4 0 1 1 15.58 19.695 50 26.0027 7 329 336 2012-12-19 19:00:00 19 2 718.0 0 0
10882 2012-12-19 20:00:00 4 0 1 1 14.76 17.425 57 15.0013 10 231 241 2012-12-19 20:00:00 20 2 718.0 0 0
10883 2012-12-19 21:00:00 4 0 1 1 13.94 15.910 61 15.0013 4 164 168 2012-12-19 21:00:00 21 2 718.0 0 0
10884 2012-12-19 22:00:00 4 0 1 1 13.94 17.425 61 6.0032 12 117 129 2012-12-19 22:00:00 22 2 718.0 0 0
10885 2012-12-19 23:00:00 4 0 1 1 13.12 16.665 66 8.9981 4 84 88 2012-12-19 23:00:00 23 2 718.0 0 0

10886 rows × 19 columns

### 从数据中,把原始的时间字段等踢掉
# remove old data features
dataRel = data.drop(['datetime', 'count','date','time','dayofweek'], axis=1)
dataRel.head()
season holiday workingday weather temp atemp humidity windspeed casual registered hour dateDays Saturday Sunday
0 1 0 0 1 9.84 14.395 81 0.0 3 13 0 0.0 1 0
1 1 0 0 1 9.02 13.635 80 0.0 8 32 1 0.0 1 0
2 1 0 0 1 9.02 13.635 80 0.0 5 27 2 0.0 1 0
3 1 0 0 1 9.84 14.395 75 0.0 3 10 3 0.0 1 0
4 1 0 0 1 9.84 14.395 75 0.0 0 1 4 0.0 1 0
### 特征向量化 我们这里打算用scikit-learn来建模。对于pandas的dataframe我们有方法/函数可以直接转成python中的dict。 另外,在这里我们要对离散值和连续值特征区分一下了,以便之后分开做不同的特征处理。
from sklearn.feature_extraction import DictVectorizer
# 我们把连续值的属性放入一个dict中
featureConCols = ['temp','atemp','humidity','windspeed','dateDays','hour']
dataFeatureCon = dataRel[featureConCols]
dataFeatureCon = dataFeatureCon.fillna( 'NA' ) #in case I missed any
X_dictCon = dataFeatureCon.T.to_dict().values() 

# 把离散值的属性放到另外一个dict中
featureCatCols = ['season','holiday','workingday','weather','Saturday', 'Sunday']
dataFeatureCat = dataRel[featureCatCols]
dataFeatureCat = dataFeatureCat.fillna( 'NA' ) #in case I missed any
X_dictCat = dataFeatureCat.T.to_dict().values() 

# 向量化特征
vec = DictVectorizer(sparse = False)
X_vec_cat = vec.fit_transform(X_dictCat)
X_vec_con = vec.fit_transform(X_dictCon)
dataFeatureCon.head()
temp atemp humidity windspeed dateDays hour
0 9.84 14.395 81 0.0 0.0 0
1 9.02 13.635 80 0.0 0.0 1
2 9.02 13.635 80 0.0 0.0 2
3 9.84 14.395 75 0.0 0.0 3
4 9.84 14.395 75 0.0 0.0 4
X_vec_con
array([[ 14.395 , 0. , 0. , 81. , 9.84 , 0. ], [ 13.635 , 0. , 1. , 80. , 9.02 , 0. ], [ 13.635 , 0. , 2. , 80. , 9.02 , 0. ], …, [ 15.91 , 718. , 21. , 61. , 13.94 , 15.0013], [ 17.425 , 718. , 22. , 61. , 13.94 , 6.0032], [ 16.665 , 718. , 23. , 66. , 13.12 , 8.9981]])
dataFeatureCat.head()
season holiday workingday weather Saturday Sunday
0 1 0 0 1 1 0
1 1 0 0 1 1 0
2 1 0 0 1 1 0
3 1 0 0 1 1 0
4 1 0 0 1 1 0
X_vec_cat
array([[ 1.,  0.,  0.,  1.,  1.,  0.],
       [ 1.,  0.,  0.,  1.,  1.,  0.],
       [ 1.,  0.,  0.,  1.,  1.,  0.],
       ..., 
       [ 0.,  0.,  0.,  4.,  1.,  1.],
       [ 0.,  0.,  0.,  4.,  1.,  1.],
       [ 0.,  0.,  0.,  4.,  1.,  1.]])

标准化连续值特征

我们要对连续值属性做一些处理,最基本的当然是标准化,让连续值属性处理过后均值为0,方差为1。
这样的数据放到模型里,对模型训练的收敛和模型的准确性都有好处

from sklearn import preprocessing
# 标准化连续值数据
scaler = preprocessing.StandardScaler().fit(X_vec_con)
X_vec_con = scaler.transform(X_vec_con)
X_vec_con
array([[-1.09273697, -1.70912256, -1.66894356,  0.99321305, -1.33366069,
        -1.56775367],
       [-1.18242083, -1.70912256, -1.52434128,  0.94124921, -1.43890721,
        -1.56775367],
       [-1.18242083, -1.70912256, -1.379739  ,  0.94124921, -1.43890721,
        -1.56775367],
       ..., 
       [-0.91395927,  1.70183906,  1.36770431, -0.04606385, -0.80742813,
         0.26970368],
       [-0.73518157,  1.70183906,  1.51230659, -0.04606385, -0.80742813,
        -0.83244247],
       [-0.82486544,  1.70183906,  1.65690887,  0.21375537, -0.91267464,
        -0.46560752]])

类别特征编码

最常用的当然是one-hot编码咯,比如颜色 红、蓝、黄 会被编码为[1, 0, 0],[0, 1, 0],[0, 0, 1]

from sklearn import preprocessing
# one-hot编码
enc = preprocessing.OneHotEncoder()
enc.fit(X_vec_cat)
X_vec_cat = enc.transform(X_vec_cat).toarray()
X_vec_cat
array([[ 1.,  0.,  0., ...,  1.,  1.,  0.],
       [ 1.,  0.,  0., ...,  1.,  1.,  0.],
       [ 1.,  0.,  0., ...,  1.,  1.,  0.],
       ..., 
       [ 0.,  1.,  1., ...,  0.,  0.,  1.],
       [ 0.,  1.,  1., ...,  0.,  0.,  1.],
       [ 0.,  1.,  1., ...,  0.,  0.,  1.]])

把特征拼一起

把离散和连续的特征都组合在一起

import numpy as np
# combine cat & con features
X_vec = np.concatenate((X_vec_con,X_vec_cat), axis=1)
X_vec
array([[-1.09273697, -1.70912256, -1.66894356, ...,  1.        ,
         1.        ,  0.        ],
       [-1.18242083, -1.70912256, -1.52434128, ...,  1.        ,
         1.        ,  0.        ],
       [-1.18242083, -1.70912256, -1.379739  , ...,  1.        ,
         1.        ,  0.        ],
       ..., 
       [-0.91395927,  1.70183906,  1.36770431, ...,  0.        ,
         0.        ,  1.        ],
       [-0.73518157,  1.70183906,  1.51230659, ...,  0.        ,
         0.        ,  1.        ],
       [-0.82486544,  1.70183906,  1.65690887, ...,  0.        ,
         0.        ,  1.        ]])

最后的特征,前6列是标准化过后的连续值特征,后面是编码后的离散值特征

对结果值也处理一下

拿到结果的浮点数值

# 对Y向量化
Y_vec_reg = dataRel['registered'].values.astype(float)
Y_vec_cas = dataRel['casual'].values.astype(float)
Y_vec_reg
array([  13.,   32.,   27., ...,  164.,  117.,   84.])
Y_vec_cas
array([  3.,   8.,   5., ...,   4.,  12.,   4.])
0
0

查看评论
* 以上用户言论只代表其个人观点,不代表CSDN网站的观点或立场
    个人资料
    • 访问:20095次
    • 积分:769
    • 等级:
    • 排名:千里之外
    • 原创:59篇
    • 转载:2篇
    • 译文:0篇
    • 评论:6条
    文章存档
    最新评论