Question

我有2 df

df1：

ID       X Y Cond
Johnson  2 3 fine
Sand     NAN NAN sick
Cooper   1 2 fine
Nelson   1 2 fine
Peterson 4 5 fine

和df2：

id2      X Y 
Magic    2 3 
Sand     2 3 
Cooper   1 2 
Dean     1 2

如果Cond =“ sick”和df [“ id”] = df [“ id2]，我想更新df1中的x值获得新的df1：

ID       X Y Cond
Johnson  2 3 fine
Sand     2 3  sick
Cooper   1 2 fine
Nelson   1 2 fine
Peterson 4 5 fine

我尝试过：

df1["x"] = np.where((df["cond"]=="sick")& (df1["id"]==df2["id2"]),df2["x"],"")

但是它不起作用。我得到这个ValueError：

ValueError: Can only compare identically-labeled Series objects

谢谢

Answer 1

首先将两个ID列都转换为索引值，以使DataFrame.loc可能匹配选定的行：

df11 = df1.set_index('ID')
df22 = df2.set_index('id2')

df11.loc[df11["Cond"]=="sick", ['X','Y']] = df22[['X','Y']]
df = df11.reset_index()
print (df)
         ID  X  Y  Cond
0   Johnson  2  3  fine
1      Sand  2  3  sick
2    Cooper  1  2  fine
3    Nelson  1  2  fine
4  Peterson  4  5  fine

Answer 2

您可以使用pandas数据框的where()方法来代替numpy的where函数。代码如下：

df1.loc[:,["X", "Y"]] = df1.loc[:,["X", "Y"]].where(df1["Cond"]!="sick",df2.loc[:,["X", "Y"]])

根据条件从另一个df获取价值

2 个答案: