HOWTO · Pandas

Pandas 刪除帶有 NaN 的行

本教程解釋了我們如何使用 DataFrame.notna()和 DataFrame.dropna()方法來刪除所有帶有 NaN 值的行。

作者 Suraj Joshi發佈於 2021-01-22更新於 2023-01-30
本頁內容

本教程解釋了我們如何使用 DataFrame.notna() 和 DataFrame.dropna() 方法刪除所有帶有 NaN 值的行。

我們將在下面的示例程式碼中使用 DataFrame。

import pandas as pd

roll_no = [501, 502, 503, 504, 505]

data = pd.DataFrame(
    {
        "Name": ["Alice", "Steven", "Neesham", "Chris", "Alice"],
        "Age": [19, None, 18, 21, None],
        "Income($)": [4000, 5000, None, 3500, None],
        "Expense($)": [3000, 2000, 2500, 25000, None],
    }
)

print(data)

輸出：

      Name   Age  Income($)  Expense($)
0    Alice  19.0     4000.0      3000.0
1   Steven   NaN     5000.0      2000.0
2  Neesham  18.0        NaN      2500.0
3    Chris  21.0     3500.0     25000.0
4    Alice   NaN        NaN         NaN

Pandas 使用 DataFrame.notna() 方法刪除帶有 NaN 的行

DataFrame.notna() 方法返回一個布林物件，其行數和列數與呼叫者 DataFrame 相同。如果元素不是 NaN，它將被對映到布林物件中的 True 值，如果元素是 NaN，它將被對映到 False 值。

import pandas as pd

roll_no = [501, 502, 503, 504, 505]

data = pd.DataFrame(
    {
        "Name": ["Alice", "Steven", "Neesham", "Chris", "Alice"],
        "Age": [19, None, 18, 21, None],
        "Income($)": [4000, 5000, None, 3500, None],
        "Expense($)": [3000, 2000, 2500, 25000, None],
    }
)
print("Initial DataFrame:")
print(data)

print("")

data = data[data["Income($)"].notna()]
print("DataFrame after removing rows with NaN value in Income Field:")
print(data)

輸出：

Initial DataFrame:
      Name   Age  Income($)  Expense($)
0    Alice  19.0     4000.0      3000.0
1   Steven   NaN     5000.0      2000.0
2  Neesham  18.0        NaN      2500.0
3    Chris  21.0     3500.0     25000.0
4    Alice   NaN        NaN         NaN

DataFrame after removing rows with NaN value in Income Field:
     Name   Age  Income($)  Expense($)
0   Alice  19.0     4000.0      3000.0
1  Steven   NaN     5000.0      2000.0
3   Chris  21.0     3500.0     25000.0

這裡，我們將 notna() 方法應用於 data 的 Income($) 列，它將返回一個系列物件，根據該列的值，有 True 或 False 值。當我們將布林物件作為索引傳遞給原始 DataFrame 時，我們只得到 Income($) 列沒有 NaN 值的行。

Pandas 使用 DataFrame.dropna() 方法只刪除所有列都是 NaN 值的行

import pandas as pd

roll_no = [501, 502, 503, 504, 505]

data = pd.DataFrame(
    {
        "Id": [621, 645, 210, 345, None],
        "Age": [19, None, 18, 21, None],
        "Income($)": [4000, 5000, None, 3500, None],
        "Expense($)": [3000, 2000, 2500, 25000, None],
    }
)
print("Initial DataFrame:")
print(data)

print("")

data = data.dropna(how="all")
print("DataFrame after removing rows with NaN value in All Columns:")
print(data)

輸出：

Initial DataFrame:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
1  645.0   NaN     5000.0      2000.0
2  210.0  18.0        NaN      2500.0
3  345.0  21.0     3500.0     25000.0
4    NaN   NaN        NaN         NaN

DataFrame after removing rows with NaN value in All Columns:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
1  645.0   NaN     5000.0      2000.0
2  210.0  18.0        NaN      2500.0
3  345.0  21.0     3500.0     25000.0

它只刪除 DataFrame 中所有欄位中含有 NaN 值的行。我們在 dropna() 方法中設定 how='all'，讓該方法只在行的所有列值都是 NaN 時才刪除行。

Pandas 使用 DataFrame.dropna() 方法僅在某一列的值為 NaN 的情況下才刪除行

import pandas as pd

roll_no = [501, 502, 503, 504, 505]

data = pd.DataFrame(
    {
        "Id": [621, 645, 210, 345, None],
        "Age": [19, None, 18, 21, None],
        "Income($)": [4000, 5000, None, 3500, None],
        "Expense($)": [3000, 2000, 2500, 25000, None],
    }
)
print("Initial DataFrame:")
print(data)

print("")

data = data.dropna(subset=["Id"])
print("DataFrame after removing rows with NaN value in Id Column:")
print(data)

輸出：

Initial DataFrame:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
1  645.0   NaN     5000.0      2000.0
2  210.0  18.0        NaN      2500.0
3  345.0  21.0     3500.0     25000.0
4    NaN   NaN        NaN         NaN

DataFrame after removing rows with NaN value in Id Column:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
1  645.0   NaN     5000.0      2000.0
2  210.0  18.0        NaN      2500.0
3  345.0  21.0     3500.0     25000.0

它將刪除 DataFrame 中所有僅在 Id 列中具有 NaN 值的列。

Pandas 使用 DataFrame.dropna() 方法刪除任意列為 NaN 值的行

import pandas as pd

roll_no = [501, 502, 503, 504, 505]

data = pd.DataFrame(
    {
        "Id": [621, 645, 210, 345, None],
        "Age": [19, None, 18, 21, None],
        "Income($)": [4000, 5000, None, 3500, None],
        "Expense($)": [3000, 2000, 2500, 25000, None],
    }
)
print("Initial DataFrame:")
print(data)

print("")

data = data.dropna()
print("DataFrame after removing rows with NaN value in any column:")
print(data)

輸出：

Initial DataFrame:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
1  645.0   NaN     5000.0      2000.0
2  210.0  18.0        NaN      2500.0
3  345.0  21.0     3500.0     25000.0
4    NaN   NaN        NaN         NaN

DataFrame after removing rows with NaN value in any column:
      Id   Age  Income($)  Expense($)
0  621.0  19.0     4000.0      3000.0
3  345.0  21.0     3500.0     25000.0

預設情況下，dropna() 方法將刪除所有至少有一個 NaN 值的行。

關於作者

Suraj Joshi

Suraj Joshi is a backend software engineer at Matrice.ai.

查看作者個人檔案
LinkedIn