如何在pandas数据框中查找具有正值和负值的行

Question

I have a dataframe like this:

     e_col   in_col     word_col      w_col
     31      9        algorithm    -0.053538
     31      9              ubc    -0.053578
     31      9              kth    -0.053595
     31      8              ubc    -0.053633
     30      8        algorithm     0.043637
     30      7             dale     0.053648
     28      6             dale     0.053671

I want to find the rows in which the same word_col got both positive and negative values in w_col .

So, For example here the output will be:

 31      9        algorithm    -0.053538
 30      8        algorithm     0.043637

Answer 1

Edit 2 : you may also use transform to avoid set_index/reset_index as follows:

m = df.w_col.lt(0).groupby(df.word_col).transform('nunique').eq(2)
df.loc[m]

Out[2768]:
   e_col  in_col   word_col     w_col
0     31       9  algorithm -0.053538
4     30       8  algorithm  0.043637

---

Edit 1 : a shorter way to create m is using nunique() as follows:

m =  df.w_col.lt(0).groupby(df.word_col).nunique().eq(2)

Original :
Do the following: create boolean mask on w_col less than 0 and groupby it by word_col . Next, call unique on each group and find any group having len = 2. Use this as the mask to indexing on df.set_index and reset_index back.

m = df.w_col.lt(0).groupby(df.word_col).unique().str.len().eq(2)
df.set_index('word_col').loc[m].reset_index()

Out[2738]:
    word_col  e_col  in_col     w_col
0  algorithm     31       9 -0.053538
1  algorithm     30       8  0.043637