pandas 初识（五）

1. 如何实现把一个属性（列）拆分成多列，产生pivot，形成向量信息，计算相关性？

例：

     class_    timestamp    count
0    10    2019-01-20 13:23:00    1
1    10    2019-01-20 13:24:00    2
2    10    2019-01-20 13:25:00    2
3    10    2019-01-20 13:26:00    1
4    10    2019-01-20 13:27:00    2

转为：

class_ 1 2 3 4 10
timestamp
2019-01-20 13:23:01 1.0 NaN NaN NaN NaN
2019-01-20 13:24:02 NaN NaN 2.0 NaN NaN
2019-01-20 13:25:03 NaN 2.0 NaN NaN NaN
2019-01-20 13:26:02 NaN NaN NaN 1.0 NaN
2019-01-20 13:27:05 NaN NaN NaN NaN 2.0

解决：

import pandas as pd
from pandas import Timestamp

info = {'class_': {0: 1, 1: 2, 2: 3, 3: 4, 4: 10},
 'timestamp': {0: Timestamp('2019-01-20 13:23:00'),
  1: Timestamp('2019-01-20 13:24:00'),
  2: Timestamp('2019-01-20 13:25:00'),
  3: Timestamp('2019-01-20 13:26:00'),
  4: Timestamp('2019-01-20 13:27:00')},
 'count': {0: 1, 1: 2, 2: 2, 3: 1, 4: 2}}
df = pd.DataFrame(info)  
# df.pivot(index='timestamp', columns="class_", values="count").fillna(0)  
df.pivot(index='timestamp', columns="class_", values="count")

 

 

2. 如何实现把一个属性的多列（属性唯一）合并成一列

例：

class_ 1 2 3 4 10
timestamp
2019-01-20 13:23:01 1.0 NaN NaN NaN NaN
2019-01-20 13:24:02 NaN NaN 2.0 NaN NaN
2019-01-20 13:25:03 NaN 2.0 NaN NaN NaN
2019-01-20 13:26:02 NaN NaN NaN 1.0 NaN
2019-01-20 13:27:05 NaN NaN NaN NaN 2.0

转为：

     class_    timestamp    count
0    10    2019-01-20 13:23:00    1
1    10    2019-01-20 13:24:00    2
2    10    2019-01-20 13:25:00    2
3    10    2019-01-20 13:26:00    1
4    10    2019-01-20 13:27:00    2

解决：

import pandas as pd
from pandas import Timestampinfo = {'class_': {0: 1, 1: 2, 2: 3, 3: 4, 4: 10},
 'timestamp': {0: Timestamp('2019-01-20 13:23:00'),
  1: Timestamp('2019-01-20 13:24:00'),
  2: Timestamp('2019-01-20 13:25:00'),
  3: Timestamp('2019-01-20 13:26:00'),
  4: Timestamp('2019-01-20 13:27:00')},
 'count': {0: 1, 1: 2, 2: 2, 3: 1, 4: 2}}
df = pd.DataFrame(info)  
# df1 = _df.pivot(index='timestamp', columns="class_", values="count").dropna()
df1 = _df.pivot(index='timestamp', columns="class_", values="count")
df1 = _df.stack().reset_index()
df1.columns = ["class_", "count"]