176 lines
5.5 KiB
Markdown
176 lines
5.5 KiB
Markdown
|
# 如何用 0 替换熊猫数据帧中的 NaN 值？
|
|||
|
|
|
|||
|
|
> 原文：<https://www.askpython.com/python-modules/pandas/replace-nan-values-with-zero>
|
|||
|
|
|
|||
|
|
在 Python 中，NaN 的意思是**而不是数字**。这表示具有 NaN 值的条目要么未定义，要么在数据集中缺失。它是浮点值，不能转换为任何其他数据类型。
|
|||
|
|
|
|||
|
|
NaN 值是不可取的，尤其是在机器学习模型中，因为它们可能导致训练不准确的模型。这些值可以由计算项代替，如平均值、中值或基于数据集的任何其他合适的值。
|
|||
|
|
|
|||
|
|
***也读:[如何用空字符串替换熊猫中的 NAN 值？](https://www.askpython.com/python-modules/pandas/replace-nan-values-with-empty-string)***
|
|||
|
|
|
|||
|
|
本教程将着眼于我们如何在 [Pandas 数据帧](https://www.askpython.com/python/pandas-dataframe-vs-numpy-arrays)中用 0 替换 NaN 值。让我们首先创建一个数据框。
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
## 创建熊猫数据框架
|
|||
|
|
|
|||
|
|
```py
|
|||
|
|
import pandas as pd
|
|||
|
|
import numpy as np
|
|||
|
|
|
|||
|
|
data = {
|
|||
|
|
'Mobile Model Number': [6, np.nan, 2, np.nan, 7, 3, 5,
|
|||
|
|
np.nan, 21, 12, np.nan],
|
|||
|
|
'Price': [30000, 5200, 6000, np.nan, np.nan, 15000, 36000,
|
|||
|
|
np.nan, 4500, np.nan, 2300],
|
|||
|
|
'Rating': [3.1, 3.0, np.nan, 4.6, np.nan, np.nan, 2.8, 4.7,
|
|||
|
|
np.nan, 3.0, np.nan]
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
df = pd.DataFrame(data)
|
|||
|
|
|
|||
|
|
df
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
Data Frame
|
|||
|
|
|
|||
|
|
上面是一个由 3 列组成的数据框:**手机型号**、**价格**和**评级**。到目前为止，所有这些列都包含一些 NaN 值。
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
## Python 函数替换 NaN 值
|
|||
|
|
|
|||
|
|
Python 中主要有两个函数可以用来替换 NaN 值:
|
|||
|
|
1。`replace()`
|
|||
|
|
2。`fillna()`
|
|||
|
|
|
|||
|
|
关于 **replace()** 函数可以参考[这篇](https://www.askpython.com/python/string/python-replace-function)了解更多，关于 **fillna()** 函数可以参考[这篇](https://www.askpython.com/python/pandas-fillna-method)文章了解更多。
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
## 用 0 替换 NaN 值的示例
|
|||
|
|
|
|||
|
|
让我们从几个替换 NaN 值的例子开始，并理解这在代码中是如何工作的。
|
|||
|
|
|
|||
|
|
### 使用 replace()函数:
|
|||
|
|
|
|||
|
|
#### a.使用 replace()将单个列中的 NaN 值替换为 0
|
|||
|
|
|
|||
|
|
```py
|
|||
|
|
data = {
|
|||
|
|
'Mobile Model Number': [6, np.nan, 2, np.nan, 7, 3, 5,
|
|||
|
|
np.nan, 21, 12, np.nan],
|
|||
|
|
'Price': [30000, 5200, 6000, np.nan, np.nan, 15000, 36000,
|
|||
|
|
np.nan, 4500, np.nan, 2300],
|
|||
|
|
'Rating': [3.1, 3.0, np.nan, 4.6, np.nan, np.nan, 2.8, 4.7,
|
|||
|
|
np.nan, 3.0, np.nan]
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# applying the replace method on a single column
|
|||
|
|
df['Rating'] = df['Rating'].replace(np.nan, 0)
|
|||
|
|
|
|||
|
|
df
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
replace(): Replace NaN in a Single Column With 0
|
|||
|
|
|
|||
|
|
在上面的代码中，我们应用了`replace()`函数，将 dataframe 的“Rating”列中的 NaN 值替换为 0。因此，该列现在用 0 代替了以前的 NaN 值。
|
|||
|
|
|
|||
|
|
#### b.使用 replace()将整个数据框中的 NaN 值替换为 0
|
|||
|
|
|
|||
|
|
```py
|
|||
|
|
data = {
|
|||
|
|
'Mobile Model Number': [6, np.nan, 2, np.nan, 7, 3, 5,
|
|||
|
|
np.nan, 21, 12, np.nan],
|
|||
|
|
'Price': [30000, 5200, 6000, np.nan, np.nan, 15000, 36000,
|
|||
|
|
np.nan, 4500, np.nan, 2300],
|
|||
|
|
'Rating': [3.1, 3.0, np.nan, 4.6, np.nan, np.nan, 2.8, 4.7,
|
|||
|
|
np.nan, 3.0, np.nan]
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# applying the replace method on the entire dataframe
|
|||
|
|
df = df.replace(np.nan, 0)
|
|||
|
|
|
|||
|
|
df
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
replace(): Replace NaN In the Entire Dataframe With 0
|
|||
|
|
|
|||
|
|
在这种情况下，我们将整个数据帧中的所有 NaN 值一次全部替换为 0。
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
### 使用 fillna()函数:
|
|||
|
|
|
|||
|
|
这个例子将使用我们之前探索过的 [fillna()函数](https://www.askpython.com/python/pandas-fillna-method)。
|
|||
|
|
|
|||
|
|
#### a.使用 fillna()将单个列中的 NaN 值替换为 0
|
|||
|
|
|
|||
|
|
```py
|
|||
|
|
data = {
|
|||
|
|
'Mobile Model Number': [6, np.nan, 2, np.nan, 7, 3, 5,
|
|||
|
|
np.nan, 21, 12, np.nan],
|
|||
|
|
'Price': [30000, 5200, 6000, np.nan, np.nan, 15000, 36000,
|
|||
|
|
np.nan, 4500, np.nan, 2300],
|
|||
|
|
'Rating': [3.1, 3.0, np.nan, 4.6, np.nan, np.nan, 2.8, 4.7,
|
|||
|
|
np.nan, 3.0, np.nan]
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# applying the fillna method on a single column
|
|||
|
|
df['Mobile Model Number'] = df['Mobile Model Number'].fillna(0)
|
|||
|
|
|
|||
|
|
df
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
fillna(): Replace NaN in a Single Column With 0
|
|||
|
|
|
|||
|
|
这里，我们将“评级”列中的所有 NaN 值替换为 0。
|
|||
|
|
|
|||
|
|
#### b.使用 fillna()将整个数据帧中的 NaN 值替换为 0
|
|||
|
|
|
|||
|
|
```py
|
|||
|
|
data = {
|
|||
|
|
'Mobile Model Number': [6, np.nan, 2, np.nan, 7, 3, 5,
|
|||
|
|
np.nan, 21, 12, np.nan],
|
|||
|
|
'Price': [30000, 5200, 6000, np.nan, np.nan, 15000, 36000,
|
|||
|
|
np.nan, 4500, np.nan, 2300],
|
|||
|
|
'Rating': [3.1, 3.0, np.nan, 4.6, np.nan, np.nan, 2.8, 4.7,
|
|||
|
|
np.nan, 3.0, np.nan]
|
|||
|
|
}
|
|||
|
|
# applying the fillna method on the entire dataframe
|
|||
|
|
df.fillna(0)
|
|||
|
|
|
|||
|
|
df
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
fillna(): Replace NaN in the Entire Dataframe With 0
|
|||
|
|
|
|||
|
|
在这种情况下，我们使用`fillna()`函数将数据帧中的所有 NaN 值一次全部替换为 0。
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
## 摘要
|
|||
|
|
|
|||
|
|
因此，我们已经看到了如何在数据帧中用 0 替换 NaN 值。要了解更多关于熊猫和其他 Python 相关概念的信息，也请查看我们的其他博客！
|
|||
|
|
|
|||
|
|
* * *
|
|||
|
|
|
|||
|
|
## 参考
|
|||
|
|
|
|||
|
|
* replace() [正式文档](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.replace.html)
|
|||
|
|
* fillna() [官方文档](https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.fillna.html)
|