一.什麼是pandas
1.1定義
是基於numpy 的一種爲了解決數據分析任務提供了高效地操作大型數據集所需的工具，
Pandas 納入了大量庫和一些標準的數據模型，提供了大量能使我們快速便捷地處理數據的函數和方法。
1.2數據結構
Series：一維數組，與Numpy中的一維array類似。二者與Python基本的數據結構List也很相近。Series如今能保存不同種數據類型，字符串、boolean值、數字等都能保存在Series中。
Time- Series：以時間爲索引的Series。
DataFrame：二維的表格型數據結構。很多功能與R中的data.frame類似。可以將DataFrame理解爲Series的容器。以下的內容主要以DataFrame爲主。
Panel ：三維的數組，可以理解爲DataFrame的容器。
二.創建
2.1創建series
s = pd.Series([1,3,5,4,6,8])
0 1
1 3
2 5
3 4
4 6
5 8
dtype: int64
2.2創建time-series
dates = pd.date_range('20181001',periods=6)
DatetimeIndex(['2018-10-01', '2018-10-02', '2018-10-03', '2018-10-04',
'2018-10-05', '2018-10-06'],
dtype='datetime64[ns]', freq='D')
2.3創建dataframe
df = pd.DataFrame(np.random.randn(6,4),index=dates,columns=list('ABCD'))
A B C D
2018-10-01 -0.419864 -0.716621 -1.522754 -0.896049
2018-10-02 0.361538 0.312017 0.427167 -1.191305
2018-10-03 -2.019418 -0.895160 0.425233 1.467052
2018-10-04 1.186440 0.251670 -0.995712 1.845424
2018-10-05 0.509954 1.543433 -0.036554 -0.426340
2018-10-06 0.585052 -1.347779 -0.115104 -0.058916
df2 = pd.DataFrame({ 'A' : 1.,
'B' : pd.Timestamp('20181002'),
'C' : pd.Series(1,index=list(range(4)),dtype='float32'),
'D' : np.array([3] * 4,dtype='int32'),
'E' : pd.Categorical(["test","train","test","train"]),
'F' : 'hello' })
A B C D E F
0 1.0 2018-10-02 1.0 3 test hello
1 1.0 2018-10-02 1.0 3 train hello
2 1.0 2018-10-02 1.0 3 test hello
3 1.0 2018-10-02 1.0 3 train hello
2.4創建panel
data = pd.Panel(np.random.rand(2,4,5))
<class 'pandas.core.panel.Panel'>
Dimensions: 2 (items) x 4 (major_axis) x 5 (minor_axis)
Items axis: 0 to 1
Major_axis axis: 0 to 3
Minor_axis axis: 0 to 4
三.入門（主要以dataframe爲主）
df：
A B C D
2018-10-01 1.097650 0.600073 1.296362 -1.119968
2018-10-02 0.931835 -1.174313 1.247509 0.462275
2018-10-03 -1.481357 1.737640 -0.853880 0.146072
2018-10-04 3.658062 -0.276350 -0.529113 0.389288
2018-10-05 -0.867468 0.153941 1.182122 1.380770
2018-10-06 -0.019635 -1.102165 0.503295 0.403443
df.head()
A B C D
2018-10-01 1.097650 0.600073 1.296362 -1.119968
2018-10-02 0.931835 -1.174313 1.247509 0.462275
2018-10-03 -1.481357 1.737640 -0.853880 0.146072
2018-10-04 3.658062 -0.276350 -0.529113 0.389288
2018-10-05 -0.867468 0.153941 1.182122 1.380770
df.tail(4)
A B C D
2018-10-03 -1.575419 0.408146 -0.727951 0.892273
2018-10-04 0.477188 -1.134372 1.354922 -0.641237
2018-10-05 -1.060002 -0.117877 -0.245136 0.048334
2018-10-06 -0.264465 0.054226 -0.381926 -0.037507
df.index
DatetimeIndex(['2018-10-01', '2018-10-02', '2018-10-03', '2018-10-04',
'2018-10-05', '2018-10-06'],
dtype='datetime64[ns]', freq='D')
df.columns
Index(['A', 'B', 'C', 'D'], dtype='object')
df.values
[[-0.01094233 0.56624115 -0.69607107 -1.41716622]
[ 1.6608257 -0.9003276 -1.21582265 0.09932746]
[ 0.18357827 -1.31061872 0.02711289 -1.25631321]
[ 1.21774381 -0.52218704 -1.00689223 2.86039582]
[-0.85700921 0.15736927 0.88435886 -1.65245358]
[ 0.74485154 0.06683108 0.31701702 -0.0537278 ]]
df.head() 最多輸出前五行
df.tail(4) 輸出後幾行
df.index 行索引
df.columns 列索引
df.values 值
df.describe()
A B C D
count 6.000000 6.000000 6.000000 6.000000
mean 0.325588 -0.381394 -0.756222 0.864435
std 0.743853 1.210628 1.682261 0.536523
min -0.890606 -2.236455 -2.670481 0.120198
25% -0.028596 -1.112915 -2.051519 0.578564
50% 0.539035 -0.068343 -0.686917 0.830762
75% 0.806438 0.422785 -0.046963 1.162358
max 1.088299 0.959387 1.847014 1.637499
快速統計摘要
count 數量
mean 平均值
std 標準差
min 最小值
25% 第一四分位數 (Q1)，又稱“較小四分位數”，等於該樣本中所有數值由小到大排列後第25%的數字。
50% 中位數
75% 同上類似
max 最大值
df.T
2018-10-01 2018-10-02 ... 2018-10-05 2018-10-06
A -1.086541 0.806482 ... 1.464247 0.413810
B -0.043731 -0.988297 ... -0.532121 -0.214636
C 0.370472 -1.854053 ... 0.669644 1.455905
D 1.057330 1.665222 ... -0.239430 -0.392194
[4 rows x 6 columns]
轉置
df.sort_index()
A B C D
2018-10-01 -0.636390 0.270001 1.482772 0.487126
2018-10-02 0.424275 2.079136 0.234442 -1.508086
2018-10-03 -1.719585 -1.404995 0.581046 -0.588827
2018-10-04 0.206164 -0.710080 0.214864 -0.245257
2018-10-05 0.157271 -1.659809 -0.689508 -0.627286
2018-10-06 0.054739 0.081013 -0.845970 0.058534
df.sort_index(ascending=False)
A B C D
2018-10-06 0.050489 1.725587 1.401500 0.109099
2018-10-05 -1.642711 -1.246953 0.031868 -0.742620
2018-10-04 0.448737 0.953092 0.072601 0.277347
2018-10-03 -0.585020 1.247202 -0.677308 -0.082281
2018-10-02 -0.441545 0.502394 -0.823207 0.077199
2018-10-01 0.320901 -0.641877 -0.592185 -0.006243
df.sort_index(axis=1)
A B C D
2018-10-01 -0.080709 -1.268852 -0.029146 1.730056
2018-10-02 0.177026 -0.095608 0.630715 1.251889
2018-10-03 -0.324091 1.135895 1.443761 -1.656423
2018-10-04 1.341659 -0.495091 -0.925834 -0.515833
2018-10-05 0.198719 1.271994 -0.315808 0.285534
2018-10-06 -0.085695 -1.108529 -0.047992 3.094784
按frame的行索引進行排序 frame.sort_index()
按frame的列索引進行排序 frame.sort_index(axis=1)
默認升序，降序可在括號加ascending=False
df.sort_index(by=['B'],ascending=False)
df.sort_values('B',ascending=False)
A B C D
2018-10-05 -1.513641 2.605290 0.487593 0.180191
2018-10-04 -0.750031 0.744077 -1.218886 1.500481
2018-10-02 1.312716 0.043054 -0.037574 -0.579429
2018-10-01 -0.206157 -0.360247 0.024310 1.851131
2018-10-03 0.522026 -0.496297 0.465798 -0.942027
2018-10-06 0.303735 -0.860363 -0.800755 1.101880
A B C D
2018-10-05 -1.513641 2.605290 0.487593 0.180191
2018-10-04 -0.750031 0.744077 -1.218886 1.500481
2018-10-02 1.312716 0.043054 -0.037574 -0.579429
2018-10-01 -0.206157 -0.360247 0.024310 1.851131
2018-10-03 0.522026 -0.496297 0.465798 -0.942027
2018-10-06 0.303735 -0.860363 -0.800755 1.101880
按值排序，可見上面兩個式子是等價的