切片系列在熊貓

Question

考慮Seriesobj：

In [50]: obj = Series(np.arange(6,10), index = ['a', 'b', 'c', 'd']) 

In [51]: obj 
Out[51]: 
a 6 
b 7 
c 8 
d 9 
dtype: int64 

我想採取的obj片，我能做到這一點在幾個方面：

In [52]: obj[1:3] 
Out[52]: 
b 7 
c 8 
dtype: int64 

In [53]: obj['b' : 'c'] 
Out[53]: 
b 7 
c 8 
dtype: int64 

現在考慮DataFrameparam_estimates_good：

In [54]: param_estimates_good 
Out[54]: 
      a   b  sigma  a_se  b_se sigma_se success 
1968 0.648508 1.803889 0.498017 0.784340 0.082366 0.529649  1 
1972 0.539485 1.733304 0.451311 1.084170 0.174030 0.677134  1 
1973 1.205704 2.054114 1.465606 0.095780 0.052851 0.090562  1 
1974 1.398968 2.105287 2.029865 0.451929 0.056154 0.428696  1 
1975 1.570900 1.877486 2.016978 0.186177 0.052413 0.183577  1 
1976 0.688932 1.651232 0.874860 0.065038 0.099080 0.055247  1 
1977 0.816918 1.949563 0.691899 0.516742 0.083973 0.385799  1 
1980 0.730454 2.569974 2.297921 0.619403 0.157950 0.439383  1 
1986 1.053362 1.770256 1.115229 0.235353 0.063867 0.202970  1 
1993 2.531327 1.235418 2.005588 0.107785 0.011513 0.060647  1 
1994 -0.759318 2.556910 0.175695 0.052099 0.078433 0.044315  1 
1998 1.007787 1.548161 0.911332 2.538235 0.040285 2.001148  1 
2000 -0.693261 1.518839 -0.290453 3.763934 1.329302 0.868444  1 
2001 0.662391 0.650003 0.854752 0.550188 0.547999 0.376354  1 
2002 0.652630 0.424864 0.524909 0.413478 0.334703 0.251172  1 
2004 -0.169553 1.290054 -0.040504 0.279700 0.093937 0.115120  1 
2005 0.146209 1.610219 -0.233461 0.171832 0.083844 0.123676  1 
2007 -0.301397 0.822584 0.309423 1.119639 0.860818 0.377673  1 
2008 1.334283 0.065856 1.704950 0.462811 0.489639 0.427041  1 
2009 2.082782 -0.727128 1.072343 0.464726 0.093574 0.472603  1 
2010 2.309353 -1.202509 0.906165 0.037950 0.080356 0.031981  1 
2013 3.490101 -2.033734 1.468027 0.251317 0.030869 0.259732  1 
2014 1.820431 -1.961015 -0.050831 0.262710 0.176057 0.266525  1 
2016 1.818855 -0.580492 0.312369 0.450659 0.065661 0.474896  1 

我利用這片形成Seriesg：

In [55]: g = param_estimates_good['a'] 

In [56]: g 
Out[56]: 
1968 0.648508 
1972 0.539485 
1973 1.205704 
1974 1.398968 
1975 1.570900 
1976 0.688932 
1977 0.816918 
1980 0.730454 
1986 1.053362 
1993 2.531327 
1994 -0.759318 
1998 1.007787 
2000 -0.693261 
2001 0.662391 
2002 0.652630 
2004 -0.169553 
2005 0.146209 
2007 -0.301397 
2008 1.334283 
2009 2.082782 
2010 2.309353 
2013 3.490101 
2014 1.820431 
2016 1.818855 
Name: a, dtype: float64 

的g是整型的Index ...

In [57]: g.index 
Out[57]: 
Int64Index([1968, 1972, 1973, 1974, 1975, 1976, 1977, 1980, 1986, 1993, 1994, 
      1998, 2000, 2001, 2002, 2004, 2005, 2007, 2008, 2009, 2010, 2013, 
      2014, 2016], 
      dtype='int64') 

...所以我嘗試以相似的方式來切g到obj：

In [58]: g[0:7] 
Out[58]: 
1968 0.648508 
1972 0.539485 
1973 1.205704 
1974 1.398968 
1975 1.570900 
1976 0.688932 
1977 0.816918 
Name: a, dtype: float64 


In [59]: g[1968 : 1977] 
Out[59]: Series([], Name: a, dtype: float64) 

爲什麼後一種方法返回一個空Series？

Answer 1

我覺得要找到1968位置行1977，因爲它的位置選擇行 - Slicing ranges [] in docs：

隨着系列，語法工程完全按照與ndarray，返回值的片和相應的標籤

它與Selection by position in docs和iloc相同。

print g.iloc[1968 : 1977] 
Series([], Name: a, dtype: float64) 

隨着loc它完美Selection By Label in docs：

print g.loc[1968 : 1977] 
1968 0.648508 
1972 0.539485 
1973 1.205704 
1974 1.398968 
1975 1.570900 
1976 0.688932 
1977 0.816918 
Name: a, dtype: float64