Python DataFrame数据填充：利用fillna与ffill实现向下/向上填充

南乔

发布时间：2026-10-13 | 7人浏览过

原创

本文介绍如何在Python Pandas中使用fillna方法结合ffill参数实现DataFrame的向下填充，以及使用bfill进行向上填充。通过构建包含姓名、日期和房产数量的示例数据，演示如何将稀疏数据转换为连续时间序列，并处理缺失值，适用于需要基于最近已知值进行数据补全的场景。

首先，我们需要构建一个基础的DataFrame对象：

In[8]: df = pd.DataFrame({'name':list('ABCDA'),'house':[1,1,2,3,3],'date':['2010-01-01','2010-06-09','2011-12-03','2011-04-05','2012-03-23']})
In[9]: df
Out[9]: 
   date house name
0 2010-01-01  1 A
1 2010-06-09  1 B
2 2011-12-03  2 C
3 2011-04-05  3 D
4 2012-03-23  3 A

接着，将date列的数据类型转换为时间格式：

In[12]: df.date = pd.to_datetime(df.date)

该数据集记录了ABCD四人的房产信息。具体而言，A在2010-01-01拥有1套房；B在2010-06-09拥有1套房；C在2011-12-03拥有2套房；D在2011-04-05拥有3套房。此外，A在2012-02-23的数据更新后，房产数量变为2套。

我们的目标是在已知姓名和时间的情况下，推算出各人名下的房产数量：

例如，对于A而言，在2010-01-01至2012-03-23期间的任意一天，其房产数量应显示为1套；而在2012-03-23之后，房产数量则应显示为3套。

我们可以利用pandas库中的fillna方法，并指定method参数为'ffill'来实现这一目标。

首先，我们创建一个从2010-01-01到2017-12-01的日期范围DataFrame：

In[14]: time_range = pd.DataFrame(
 pd.date_range('2010-01-01','2017-12-01',freq='D'), columns=['date']).set_index("date")
In[15]: time_range
Out[15]: 
Empty DataFrame
Columns: []
Index: [2010-01-01 00:00:00, 2010-01-02 00:00:00, 2010-01-03 00:00:00, 2010-01-04 00:00:00, 2010-01-05 00:00:00, 2010-01-06 00:00:00, 2010-01-07 00:00:00, 2010-01-08 00:00:00, 2010-01-09 00:00:00, 2010-01-10 00:00:00, 2010-01-11 00:00:00, 2010-01-12 00:00:00, 2010-01-13 00:00:00, 2010-01-14 00:00:00, 2010-01-15 00:00:00, 2010-01-16 00:00:00, 2010-01-17 00:00:00, 2010-01-18 00:00:00, 2010-01-19 00:00:00, 2010-01-20 00:00:00, 2010-01-21 00:00:00, 2010-01-22 00:00:00, 2010-01-23 00:00:00, 2010-01-24 00:00:00, 2010-01-25 00:00:00, 2010-01-26 00:00:00, 2010-01-27 00:00:00, 2010-01-28 00:00:00, 2010-01-29 00:00:00, 2010-01-30 00:00:00, 2010-01-31 00:00:00, 2010-02-01 00:00:00, 2010-02-02 00:00:00, 2010-02-03 00:00:00, 2010-02-04 00:00:00, 2010-02-05 00:00:00, 2010-02-06 00:00:00, 2010-02-07 00:00:00, 2010-02-08 00:00:00, 2010-02-09 00:00:00, 2010-02-10 00:00:00, 2010-02-11 00:00:00, 2010-02-12 00:00:00, 2010-02-13 00:00:00, 2010-02-14 00:00:00, 2010-02-15 00:00:00, 2010-02-16 00:00:00, 2010-02-17 00:00:00, 2010-02-18 00:00:00, 2010-02-19 00:00:00, 2010-02-20 00:00:00, 2010-02-21 00:00:00, 2010-02-22 00:00:00, 2010-02-23 00:00:00, 2010-02-24 00:00:00, 2010-02-25 00:00:00, 2010-02-26 00:00:00, 2010-02-27 00:00:00, 2010-02-28 00:00:00, 2010-03-01 00:00:00, 2010-03-02 00:00:00, 2010-03-03 00:00:00, 2010-03-04 00:00:00, 2010-03-05 00:00:00, 2010-03-06 00:00:00, 2010-03-07 00:00:00, 2010-03-08 00:00:00, 2010-03-09 00:00:00, 2010-03-10 00:00:00, 2010-03-11 00:00:00, 2010-03-12 00:00:00, 2010-03-13 00:00:00, 2010-03-14 00:00:00, 2010-03-15 00:00:00, 2010-03-16 00:00:00, 2010-03-17 00:00:00, 2010-03-18 00:00:00, 2010-03-19 00:00:00, 2010-03-20 00:00:00, 2010-03-21 00:00:00, 2010-03-22 00:00:00, 2010-03-23 00:00:00, 2010-03-24 00:00:00, 2010-03-25 00:00:00, 2010-03-26 00:00:00, 2010-03-27 00:00:00, 2010-03-28 00:00:00, 2010-03-29 00:00:00, 2010-03-30 00:00:00, 2010-03-31 00:00:00, 2010-04-01 00:00:00, 2010-04-02 00:00:00, 2010-04-03 00:00:00, 2010-04-04 00:00:00, 2010-04-05 00:00:00, 2010-04-06 00:00:00, 2010-04-07 00:00:00, 2010-04-08 00:00:00, 2010-04-09 00:00:00, 2010-04-10 00:00:00, ...]
 
[2892 rows x 0 columns]

随后，结合前文提到的pivot_table函数对原始df进行转换，并与time_range执行merge操作。

In[16]: df = pd.pivot_table(df, columns='name', index='date')
 
In[17]: df
Out[17]: 
   house    
name   A B C D
date       
2010-01-01 1.0 NaN NaN NaN
2010-06-09 NaN 1.0 NaN NaN
2011-04-05 NaN NaN NaN 3.0
2011-12-03 NaN NaN 2.0 NaN
2012-03-23 3.0 NaN NaN NaN
In[18]: df = df.merge(time_range,how="right", left_index=True, right_index=True)

接下来，执行向下填充操作：