为什么我的原始数据框也发生了变化？

Question

对于我正在使用的数据集，它可以在 Kaggle 上的 link 上找到

我正在这样做：

import pandas as pd

df = pd.read_csv('./survey_results_public.csv')

df = df.dropna(subset=['Salary'], axis = 0).drop(['Respondent','ExpectedSalary','Salary'], axis = 1)

print(df['HoursPerWeek'].mean())
print(sum(df['HoursPerWeek'].isnull()))

# Method 1
df1 = df
df1 = df1.select_dtypes(include=['float']).fillna(df1.mean())

print(df['HoursPerWeek'].mean())
print(sum(df['HoursPerWeek'].isnull()))
print(df1['HoursPerWeek'].mean())
print(sum(df1['HoursPerWeek'].isnull()))


# Method 2
df2 = df
num_vars = df2.select_dtypes(include = ['float']).columns
for col in num_vars:
    df2[col].fillna(df2[col].mean(),inplace = True)

print(df['HoursPerWeek'].mean())
print(sum(df['HoursPerWeek'].isnull()))
print(df2['HoursPerWeek'].mean())
print(sum(df2['HoursPerWeek'].isnull()))

我的问题是：为什么“方法 2”也会改变 df，正如在最后 4 个打印语句中观察到的那样，其中空值的平均值和数量是 df 和 df2 之间的值？

当我在 python 中对普通变量做类似的事情时，这不会发生

a=2
b=a
c=a

print(a,b,c)

b += 2

print(a,b,c)

c += 3

print(a,b,c)

在这个例子中，a 没有改变。

Answer 1

您要做的是复制数据框：

...
# Method 1
df1 = df.copy()
df1 = df1.select_dtypes(include=['float']).fillna(df1.mean())
....
# Method 2
df2 = df.copy()
num_vars = df2.select_dtypes(include = ['float']).columns
...

希望这会有所帮助：D

一个很好的例子是列表：

a = [1,2,3]
b = a
a.append(4)
print("b is",b)
# output is 'b is [1,2,3,4]