Pandas4_DataFream运算

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

date1 = pd.date_range('2020-7-1', '2020-7-20')
col = ['open', 'high', 'close', 'low', 'volume',
       'price_change', 'p_change', 'turnover']
data = pd.DataFrame(np.random.randn(20, 8), index=date1, columns=col)
data

	open	high	close	low	volume	price_change	p_change	turnover
2020-07-01	-0.322513	0.267763	-1.093117	0.027041	0.509584	0.984490	-0.190861	0.517801
2020-07-02	2.089734	-0.983815	-1.046229	-0.036239	-0.256708	0.356689	-0.252771	-0.497880
2020-07-03	-2.133904	0.679809	1.807282	-0.566753	-1.837010	0.040434	-1.059297	0.555953
2020-07-04	-0.386285	0.956894	2.315036	-1.346728	-0.896410	-0.009206	0.474799	-0.496008
2020-07-05	0.751076	-1.149339	0.197884	-0.831805	-0.496330	-0.952271	-1.532102	0.425145
2020-07-06	1.756448	-0.564414	-0.750158	-1.222868	0.757163	-0.606473	0.166400	-0.793353
2020-07-07	-1.470281	0.624951	-1.186256	-0.688941	-0.746321	-1.896055	-1.641859	1.192342
2020-07-08	0.128879	-0.416039	-0.657877	-1.313704	0.945771	-2.131931	0.356988	-2.770012
2020-07-09	-1.380173	0.650889	0.033093	1.108484	2.143283	-1.000914	1.117865	0.440897
2020-07-10	0.273836	-0.454474	-0.287727	0.992828	-0.411735	-1.074596	-1.815277	-0.248398
2020-07-11	0.841820	-0.119436	0.446986	-0.290631	1.690281	-0.706691	1.628919	0.941540
2020-07-12	0.281178	1.508803	-2.302384	0.501423	-0.071525	0.060684	0.842450	1.369522
2020-07-13	-0.155110	-0.459593	-1.896780	0.860048	-0.023542	-0.524829	-0.672771	0.486148
2020-07-14	-0.687383	0.611564	0.170420	-2.491876	-0.971182	-0.705079	-1.247554	1.782681
2020-07-15	0.684804	0.765028	1.264321	1.250718	-0.833911	0.332943	1.901829	-0.623410
2020-07-16	-0.758880	-1.139305	-0.064042	-0.591721	-1.318660	-1.263623	1.448318	-0.238409
2020-07-17	-0.628691	0.481645	-0.221320	0.719116	0.459349	-0.842811	0.739307	-1.204500
2020-07-18	-1.424206	0.536644	0.513619	0.211149	-0.232468	0.230522	0.873776	-2.385903
2020-07-19	0.313115	2.312191	0.211485	0.320764	0.772857	-0.019631	-2.276235	0.401747
2020-07-20	0.372503	0.236578	0.916577	-1.189288	-0.210505	-0.370926	1.478682	-0.399276

算数运算

• add(other)
• sub(other)

data['open'].add(10).head()

2020-07-01     9.677487
2020-07-02    12.089734
2020-07-03     7.866096
2020-07-04     9.613715
2020-07-05    10.751076
Freq: D, Name: open, dtype: float64

data['open'].sub(10).head()

2020-07-01   -10.322513
2020-07-02    -7.910266
2020-07-03   -12.133904
2020-07-04   -10.386285
2020-07-05    -9.248924
Freq: D, Name: open, dtype: float64

逻辑运算

逻辑运算符号> < | &

# 例如筛选data["open"] > 0的日期数据
data["open"] > 0

2020-07-01    False
2020-07-02     True
2020-07-03    False
2020-07-04    False
2020-07-05     True
2020-07-06     True
2020-07-07    False
2020-07-08     True
2020-07-09    False
2020-07-10     True
2020-07-11     True
2020-07-12     True
2020-07-13    False
2020-07-14    False
2020-07-15     True
2020-07-16    False
2020-07-17    False
2020-07-18    False
2020-07-19     True
2020-07-20     True
Freq: D, Name: open, dtype: bool

# 逻辑判断的结果可以作为筛选的依据
data[data["open"] > 0].head()

	open	high	close	low	volume	price_change	p_change	turnover
2020-07-02	2.089734	-0.983815	-1.046229	-0.036239	-0.256708	0.356689	-0.252771	-0.497880
2020-07-05	0.751076	-1.149339	0.197884	-0.831805	-0.496330	-0.952271	-1.532102	0.425145
2020-07-06	1.756448	-0.564414	-0.750158	-1.222868	0.757163	-0.606473	0.166400	-0.793353
2020-07-08	0.128879	-0.416039	-0.657877	-1.313704	0.945771	-2.131931	0.356988	-2.770012
2020-07-10	0.273836	-0.454474	-0.287727	0.992828	-0.411735	-1.074596	-1.815277	-0.248398

# & 注意优先级 需要加（）
data[(data["high"]>0) & (data["high"]<1)].head()

	open	high	close	low	volume	price_change	p_change	turnover
2020-07-01	1	0.267763	-1.093117	0.027041	0.509584	0.984490	-0.190861	0.517801
2020-07-03	1	0.679809	1.807282	-0.566753	-1.837010	0.040434	-1.059297	0.555953
2020-07-04	1	0.956894	2.315036	-1.346728	-0.896410	-0.009206	0.474799	-0.496008
2020-07-07	1	0.624951	-1.186256	-0.688941	-0.746321	-1.896055	-1.641859	1.192342
2020-07-09	1	0.650889	0.033093	1.108484	2.143283	-1.000914	1.117865	0.440897

逻辑运算函数

• query(expr)
• expr:查询字符串

# 通过query使得刚才的过程更加方便简单
data.query('high>0 & high<1').head()

	open	high	close	low	volume	price_change	p_change	turnover
2020-07-01	1	0.267763	-1.093117	0.027041	0.509584	0.984490	-0.190861	0.517801
2020-07-03	1	0.679809	1.807282	-0.566753	-1.837010	0.040434	-1.059297	0.555953
2020-07-04	1	0.956894	2.315036	-1.346728	-0.896410	-0.009206	0.474799	-0.496008
2020-07-07	1	0.624951	-1.186256	-0.688941	-0.746321	-1.896055	-1.641859	1.192342
2020-07-09	1	0.650889	0.033093	1.108484	2.143283	-1.000914	1.117865	0.440897

• isin(列表名或[值])

# 可以判断指定值是否在所找的数据中，data['open']指定要查找的数据
data.open = 1
data[data['open'].isin([1,0,5])].head()

	open	high	close	low	volume	price_change	p_change	turnover
2020-07-01	1	0.267763	-1.093117	0.027041	0.509584	0.984490	-0.190861	0.517801
2020-07-02	1	-0.983815	-1.046229	-0.036239	-0.256708	0.356689	-0.252771	-0.497880
2020-07-03	1	0.679809	1.807282	-0.566753	-1.837010	0.040434	-1.059297	0.555953
2020-07-04	1	0.956894	2.315036	-1.346728	-0.896410	-0.009206	0.474799	-0.496008
2020-07-05	1	-1.149339	0.197884	-0.831805	-0.496330	-0.952271	-1.532102	0.425145

统计运算

describe

• 综合分析: 能够直接得出很多统计结果,count, mean, std, min, max 等

data.describe()

	open	high	close	low	volume	price_change	p_change	turnover
count	20.0	20.000000	20.000000	20.000000	20.000000	20.000000	20.000000	20.000000
mean	1.0	0.217317	-0.081459	-0.228949	-0.051401	-0.504964	0.017030	-0.077169
std	0.0	0.888158	1.152481	0.999190	0.990270	0.776446	1.276293	1.151633
min	1.0	-1.149339	-2.302384	-2.491876	-1.837010	-2.131931	-2.276235	-2.770012
25%	1.0	-0.455754	-0.824176	-0.921176	-0.768218	-0.964431	-1.106361	-0.529262
50%	1.0	0.374704	-0.015474	-0.163435	-0.221487	-0.565651	0.261694	0.081669
75%	1.0	0.658119	0.463645	0.555847	0.571479	0.045497	0.934798	0.527339
max	1.0	2.312191	2.315036	1.250718	2.143283	0.984490	1.901829	1.782681

# 25% 50% 75% 四分位数

统计函数

• count sum mean median(中位数) min max mode(众数) prod(积)

• abs std(标准差) var(方差) idxmax(最大值位置) idxmin（最小值位置）

• 对于单个函数去进行统计的时候，坐标轴还是按照默认列“columns” (axis=0, default)，

• 如果要对行“index” 需要指定(axis=1)

• max() min()

data.max()
data.min()

open            1.000000
high           -1.149339
close          -2.302384
low            -2.491876
volume         -1.837010
price_change   -2.131931
p_change       -2.276235
turnover       -2.770012
dtype: float64

• std var

data.std()

open            0.000000
high            0.888158
close           1.152481
low             0.999190
volume          0.990270
price_change    0.776446
p_change        1.276293
turnover        1.151633
dtype: float64

• idxmax()、idxmin()

data.idxmax()

open           2020-07-01
high           2020-07-19
close          2020-07-04
low            2020-07-15
volume         2020-07-09
price_change   2020-07-01
p_change       2020-07-15
turnover       2020-07-14
dtype: datetime64[ns]

data.idxmin()

open           2020-07-01
high           2020-07-05
close          2020-07-12
low            2020-07-14
volume         2020-07-03
price_change   2020-07-08
p_change       2020-07-19
turnover       2020-07-08
dtype: datetime64[ns]

累计统计函数

• cumsum 计算前1/2/3/…/n个数的和
• cummax 计算前1/2/3/…/n个数的最大值
• cummin 计算前1/2/3/…/n个数的最小值
• cumprod 计算前1/2/3/…/n个数的积

data = data.sort_index()
data2 = data['low']
data2.cumsum().plot()
plt.show()

自定义运算

• apply(func, axis=0)
• func:自定义函数
• axis=0:默认是列，axis=1为行进行运算

# 匿名函数，最大值-最小值的函数
data[['close']].apply(lambda x: x.max() - x.min(), axis=0)

close    4.61742
dtype: float64