訪問最後一個非空值的大熊貓

Question

我想用給定組的最後一個有效值填充數據幀NaNs。例如：

import pandas as pd 
import random as randy 
import numpy as np 

df_size = int(1e1)     
df = pd.DataFrame({'category': randy.sample(np.repeat(['Strawberry','Apple',],df_size),df_size), 'values': randy.sample(np.repeat([np.NaN,0,1],df_size),df_size)}, index=randy.sample(np.arange(0,10),df_size)).sort_index(by=['category'], ascending=[True]) 

提供：

 category value 
7  Apple  NaN 
6  Apple  1 
4  Apple  0 
5  Apple  NaN 
1  Apple  NaN 
0 Strawberry  1 
8 Strawberry  NaN 
2 Strawberry  0 
3 Strawberry  0 
9 Strawberry  NaN 

和列我要計算這樣的容貌：

 category value last_value 
7  Apple  NaN   NaN 
6  Apple  1   NaN 
4  Apple  0   1 
5  Apple  NaN   0 
1  Apple  NaN   0 
0 Strawberry  1   NaN 
8 Strawberry  NaN   1 
2 Strawberry  0   1 
3 Strawberry  0   0 
9 Strawberry  NaN   0 

嘗試shift()和iterrows()但無濟於事。

Answer 1

它看起來像你想先做一個ffill然後做一個shift：

In [11]: df['value'].ffill() 
Out[11]: 
7 NaN 
6  1 
4  0 
5  0 
1  0 
0  1 
8  1 
2  0 
3  0 
9  0 
Name: value, dtype: float64 

In [12]: df['value'].ffill().shift(1) 
Out[12]: 
7 NaN 
6 NaN 
4  1 
5  0 
1  0 
0  0 
8  1 
2  1 
3  0 
9  0 
Name: value, dtype: float64 

要做到這一點對每個組你必須先GROUPBY類別，然後再應用此功能：

In [13]: g = df.groupby('category') 

In [14]: g['value'].apply(lambda x: x.ffill().shift(1)) 
Out[14]: 
7 NaN 
6 NaN 
4  1 
5  0 
1  0 
0 NaN 
8  1 
2  1 
3  0 
9  0 
dtype: float64 

In [15]: df['last_value'] = g['value'].apply(lambda x: x.ffill().shift(1))