pandas 面试题挑战三

11 把数据进行cut操作

现有数据ages如下

ages = [20, 22, 25, 27, 21, 23, 37, 31, 61, 45, 41, 32, 101]

想按照[(18, 25] < (25, 35] < (35, 60] < (60, 100]]把该数据进行Categories 操作

解决办法：

bins = [18, 25, 35, 60, 100]
cats = pd.cut(ages, bins)
cats

输出

[(18, 25], (18, 25], (18, 25], (25, 35], (18, 25], ..., (60.0, 100.0], (35.0, 60.0], (35.0, 60.0], (25.0, 35.0], NaN]
Length: 13
Categories (4, interval[int64]): [(18, 25] < (25, 35] < (35, 60] < (60, 100]]

此时的cats是非常特殊的一类数据在pd中称为是Categories
Categories可以方便的通过pd.Series()转换为Series

ser_cats = pd.Series(cats)
ser_cats

输出：

0      (18.0, 25.0]
1      (18.0, 25.0]
2      (18.0, 25.0]
3      (25.0, 35.0]
4      (18.0, 25.0]
5      (18.0, 25.0]
6      (35.0, 60.0]
7      (25.0, 35.0]
8     (60.0, 100.0]
9      (35.0, 60.0]
10     (35.0, 60.0]
11     (25.0, 35.0]
12              NaN
dtype: category
Categories (4, interval[int64]): [(18, 25] < (25, 35] < (35, 60] < (60, 100]]

当然这个分类方式看起来不太符合人类的习惯。你可以通过设定labels让输出更加符合人类的习惯

group_names = ['Youth', 'YoungAdult', 'MiddleAged', 'Senior']
cats = pd.cut(ages, bins, labels=group_names)
ser_cats = pd.Series(cats)
ser_cats

输出

0          Youth
1          Youth
2          Youth
3     YoungAdult
4          Youth
5          Youth
6     MiddleAged
7     YoungAdult
8         Senior
9     MiddleAged
10    MiddleAged
11    YoungAdult
12           NaN
dtype: category
Categories (4, object): [Youth < YoungAdult < MiddleAged < Senior]

12 把数据进行qcut操作

现有数据ages如下：

ages = [20, 22, 25, 27, 21, 23, 37, 31, 61, 45, 41, 32, 101]

要把数据按照[(19.999, 23.0] < (23.0, 31.0] < (31.0, 41.0] < (41.0, 101.0]]分成4类
也就是把数据的按照百分位[0, .25, .5, .75, 1.]进行分类

解决方法如下：

ages = [20, 22, 25, 27, 21, 23, 37, 31, 61, 45, 41, 32, 101]
cats = pd.qcut(ages, 4) 
#按[0, .25, .5, .75, 1.]分类
cats

输出

[(19.999, 23.0], (19.999, 23.0], (23.0, 31.0], (23.0, 31.0], (19.999, 23.0], ..., (41.0, 101.0], (41.0, 101.0], (31.0, 41.0], (31.0, 41.0], (41.0, 101.0]]
Length: 13
Categories (4, interval[float64]): [(19.999, 23.0] < (23.0, 31.0] < (31.0, 41.0] < (41.0, 101.0]]

参考pandas的cut&qcut函數