如何将 Pandas 列转换为 for 循环的两倍？

Question

提问by Cody

I have a pandas dataframe called df10.

我有一个名为 .pandas 的数据框df10。

cols10 = list(df10.columns)

I want to change the data type to doublewhere the column name starts with "m_"

我想将数据类型更改double为列名开头的位置"m_"

for param in cols10:
    if param.startswith("m_"):
        df10[[param]] = df10[[param]].apply(pd.to_double)

However, I'm getting this error:

但是，我收到此错误：

AttributeError: 'module' object has no attribute 'to_double'

Answer 1

the function to_doubledoesn't exist in pandas. pandas datatypes are essentially numpy data types

该功能to_double在Pandas中不存在。pandas 数据类型本质上是numpy 数据类型

i'm assuming you mean float64 by double

我假设你的意思是 float64 by double

you can either let numpy decide the precision for you

你可以让 numpy 为你决定精度

for col in cols10:
    if col.startswith('m_'):
        df[col] = df[col].astype(float)

or specify the precision yourself

或自己指定精度

for col in cols10:
    if col.startswith('m_'):
        df[col] = df[col].astype(np.float64) # or np.float32 or np.float16

Answer 2

You can do this without loop,

你可以在没有循环的情况下做到这一点，

reqd_cols = df10.columns[df10.columns.str.startswith('m_')]
df10[reqd_cols] = df10[reqd_cols].astype('float64')

Answer 3

Use the astype()function to convert the column to float(python doesn't have a doubletype like C).

使用该astype()函数将列转换为float（python 没有double像 C这样的类型）。

cols10 = [c for c in df10 if c.startswith("m_")]
df10.loc[:, cols10 ] = df10.loc[:, cols10].astype(float)

This approach is faster than using a forloop, but if you insist on looping over the columns:

这种方法比使用for循环更快，但如果您坚持循环遍历列：

cols10 = [c for c in df10 if c.startswith("m_")]
for param in cols10:
    df10[param] = df10[param].astype(float)