pandas操作

python中使用了pandas的一些操作，特此记录下来：

生成DataFrame

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_2'],
    'label': ["a,b", 'e,f,g'],
})
print(data)

得到结果为：

   label v_id
0    a,b  v_1
1  e,f,g  v_2

按照逗号分隔并拼接

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_2'],
    'label': ["a,b", 'e,f,g'],
})
df = data.drop('label', axis=1).join(data['label'].str.split(',', expand=True).stack().reset_index(level=1, drop=True).rename('label'))
print(df)

得到结果为：

  v_id label
0  v_1     a
0  v_1     b
1  v_2     e
1  v_2     f
1  v_2     g

筛选符合条件的行

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
})
target_label = data.loc[data['label'].isin(["e", "f"])]
print(target_label)

得到结果为：

  v_id label
1  v_2     e
1  v_2     f

筛选不符合条件的行

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
    'num': [1, 2, 3, 4, 5],
})

other_label1 = data[~data['label'].isin(["f", "g"])]
print(other_label1)

other_label2 = data.query("num<=3 & label!='b'")
print(other_label2)

得到结果为：

  v_id label
0  v_1     a
0  v_1     b
1  v_2     e

  label  num v_id
0     a    1  v_1
2     e    3  v_2

替换某一列的值

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
})

df = data.copy()
df.loc[df["label"] != "", 'label'] = "1"
print(df)

得到结果为：

  v_id label
0  v_1     1
0  v_1     1
1  v_2     1
1  v_2     1
1  v_2     1

取某一列转换成list

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
})
print(data["label"].values.tolist())

得到结果为：

['a', 'b', 'e', 'f', 'g']

按照某一列去重

import pandas as pd

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
})
print(data.drop_duplicates(subset=['v_id']))

得到结果为：

  v_id label
0  v_1     a
1  v_2     e

复制dataframe并拼接

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
})

data_copy = data.copy()
times = 2
for i in range(times):
    data_copy = pd.concat([data_copy,data])
print(data_copy)

得到结果为：

  v_id label
0  v_1     a
0  v_1     b
1  v_2     e
1  v_2     f
1  v_2     g
0  v_1     a
0  v_1     b
1  v_2     e
1  v_2     f
1  v_2     g
0  v_1     a
0  v_1     b
1  v_2     e
1  v_2     f
1  v_2     g

更改某一列类型

data = pd.DataFrame({
    'v_id': ["v_1", 'v_1', "v_2", "v_2","v_2"],
    'label': ["a", 'b', "e", "f", "g"],
    'num': [1.0, 2.0, 3.0, 4.0, 5.0],
})
data["num"] = data[["num"]].astype(int)
print(data)

得到结果为：

  label  num v_id
0     a    1  v_1
1     b    2  v_1
2     e    3  v_2
3     f    4  v_2
4     g    5  v_2