运算

import pandas as pd
import numpy as np

Dataframe运算

df = pd.DataFrame(np.random.rand(4, 3), index=['a', 'b', 'c', 'd'], columns=['first', 'second', 'third'])
df

	first	second	third
a	0.753379	0.518270	0.843628
b	0.415295	0.641808	0.183964
c	0.458310	0.527492	0.628082
d	0.123897	0.896220	0.021434

# point-wise
np.exp(df)

	first	second	third
a	2.124166	1.679120	2.324786
b	1.514818	1.899914	1.201973
c	1.581400	1.694677	1.874012
d	1.131899	2.450325	1.021666

列间运算

df['minus'] = df['first'] - df['second']
df

	first	second	third	minus
a	0.753379	0.518270	0.843628	0.235109
b	0.415295	0.641808	0.183964	-0.226513
c	0.458310	0.527492	0.628082	-0.069182
d	0.123897	0.896220	0.021434	-0.772324

df['quadratic'] = df['third'] ** 2
df

	first	second	third	minus	quadratic
a	0.753379	0.518270	0.843628	0.235109	0.711708
b	0.415295	0.641808	0.183964	-0.226513	0.033843
c	0.458310	0.527492	0.628082	-0.069182	0.394487
d	0.123897	0.896220	0.021434	-0.772324	0.000459

# 使用drop函数删列、行
df.drop(columns=['minus', 'quadratic'], inplace=True)
df

	first	second	third
a	0.753379	0.518270	0.843628
b	0.415295	0.641808	0.183964
c	0.458310	0.527492	0.628082
d	0.123897	0.896220	0.021434

replace和apply函数

df['third'] = 1
# 用100代替1
df.replace(1, 100, inplace=True)
df

	first	second	third
a	0.753379	0.518270	100
b	0.415295	0.641808	100
c	0.458310	0.527492	100
d	0.123897	0.896220	100

# 每个元素开平方
df.apply(lambda x: x ** 0.5)

	first	second	third
a	0.867974	0.719910	10.0
b	0.644434	0.801130	10.0
c	0.676986	0.726287	10.0
d	0.351990	0.946689	10.0

# 统计各列的和
df.apply(np.sum, axis=1)

a    101.271649
b    101.057103
c    100.985803
d    101.020117
dtype: float64

合并Dataframe

df1 = pd.DataFrame(np.random.randint(0, 10, (2, 3)), index=np.arange(2), columns=['a', 'b', 'c'])
df1

	a	b	c
0	5	8	6
1	8	6	2

df2 = pd.DataFrame(np.random.randint(5, 15, (3, 3)), index=np.arange(4, 7), columns=['a', 'b', 'c'])
df2

	a	b	c
4	6	14	11
5	9	11	13
6	9	5	6

# concat合并
pd.concat([df1, df2])

	a	b	c
0	5	8	6
1	8	6	2
4	6	14	11
5	9	11	13
6	9	5	6