Pandas分箱大法：让连续数据秒变‘箱’中豪杰，数据界的‘变形金刚’来袭！

1. 分箱操作

• 嘿，小伙伴们，你们知道吗？分箱操作，简直就是数据界的魔术师！想象一下，那些密密麻麻、让人眼花的连续数字，经过这位大师的手，瞬间变成了井然有序的小盒子——每个盒子都装着一段数字，就像是把糖果按颜色分好一样。这样一来，原本复杂的数据变得简单易懂，异常值也乖乖就范，不再捣乱。咱们的数据分析、建模之路，从此变得顺畅无阻，简直就是数据处理的“一键美颜”啊！
• 接着咱们聊聊数据界的两大“分箱高手”——等距分箱与等频分箱！等距分箱，就像是给数据量身高，不管胖瘦高矮，一刀切下去，大家间隔都相等，整整齐齐排排站。而等频分箱呢，它更像个公平分配的小能手，确保每个箱子里数据的数量都差不多，就像是分蛋糕，力求人人有份，公平公正！这两大高手各有千秋，让数据处理变得既有趣又高效，你们说是不是啊？

2. 导包与数据准备

# 导包
import numpy as np
import pandas as pd

# 创建数据
data = np.random.randint(0,100,size=(5,3))
df = pd.DataFrame(data=data,columns=["Python","NumPy","Pandas"])
df

	Python	NumPy	Pandas
0	95	59	16
1	81	68	63
2	73	70	35
3	30	4	47
4	58	18	80

3. 等距分箱

s = pd.cut(df.Python,bins=4)
s

0      (78.75, 95.0]
1      (78.75, 95.0]
2      (62.5, 78.75]
3    (29.935, 46.25]
4      (46.25, 62.5]
Name: Python, dtype: category
Categories (4, interval[float64, right]): [(29.935, 46.25] < (46.25, 62.5] < (62.5, 78.75] < (78.75, 95.0]]

s.value_counts()

Python
(78.75, 95.0]      2
(29.935, 46.25]    1
(46.25, 62.5]      1
(62.5, 78.75]      1
Name: count, dtype: int64

s.value_counts().plot.bar()

<Axes: xlabel='Python'>

pd.cut(
    df.Python,   # 分箱数据
    bins=[0,30,60,80,100],   # 分箱断点
    right=False,   # 左闭右开，默认是左开右闭
    labels=["D","C","B","A"]   # 分箱后分类的标签
)

0    A
1    A
2    B
3    C
4    C
Name: Python, dtype: category
Categories (4, object): ['D' < 'C' < 'B' < 'A']

4. 等频分箱

pd.qcut(
    df.Python,   # 分箱数据
    q=4,   # 4等分
)

0      (81.0, 95.0]
1      (73.0, 81.0]
2      (58.0, 73.0]
3    (29.999, 58.0]
4    (29.999, 58.0]
Name: Python, dtype: category
Categories (4, interval[float64, right]): [(29.999, 58.0] < (58.0, 73.0] < (73.0, 81.0] < (81.0, 95.0]]

pd.qcut(
    df.Python,   # 分箱数据
    q=4,   # 4等分
    labels=["D","C","B","A"]   # 分箱后分类的标签
)

0    A
1    B
2    C
3    D
4    D
Name: Python, dtype: category
Categories (4, object): ['D' < 'C' < 'B' < 'A']