5. 快速选取标量

  1. #  通过将行标签赋值给一个变量，用loc选取
  2.  In[37]: college = pd.read_csv('data/college.csv', index_col='INSTNM')
  3.          cn = 'Texas A & M University-College Station'
  4.          college.loc[cn, 'UGDS_WHITE']
  5. Out[37]: 0.66099999999999992
  1. #  at可以实现同样的功能
  2.  In[38]: college.at[cn, 'UGDS_WHITE']
  3. Out[38]: 0.66099999999999992
  1. #  用魔术方法%timeit，对速度进行比较
  2.  In[39]: %timeit college.loc[cn, 'UGDS_WHITE']
  3. Out[39]: 9.93 µs ± 274 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
  4.  In[40]: %timeit college.at[cn, 'UGDS_WHITE']
  5. Out[40]: 6.69 µs ± 223 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

.iat和.at只接收标量值，是专门用来取代.iloc和.loc选取标量的，可以节省大概2.5微秒。

  1. #  用get_loc找到整数位置，再进行速度比较
  2.  In[41]: row_num = college.index.get_loc(cn)
  3.          col_num = college.columns.get_loc('UGDS_WHITE')
  4.  In[42]: row_num, col_num
  5. Out[42]: (3765, 10)
  7.  In[43]: %timeit college.iloc[row_num, col_num]
  8. Out[43]: 11.1 µs ± 426 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
  10.  In[44]: %timeit college.iat[row_num, col_num]
  11. Out[44]: 7.47 µs ± 109 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
  13.  In[45]: %timeit college.iloc[5, col_num]
  14. Out[45]: 10.8 µs ± 467 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
  16.  In[46]: %timeit college.iat[5, col_num]
  17. Out[46]: 7.12 µs ± 297 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

更多

  1. #  Series对象也可以使用.iat和.at选取标量
  2.  In[47]: state = college['STABBR']
  3.  In[48]: state.iat[1000]
  4. Out[48]: 'IL'
  6.  In[49]: state.at['Stanford University']
  7. Out[49]: 'CA'