Shifting and reverting multiple rows in pandas dataframe

Question

I have the following dataframe and wish to shift over the 0 values to the right and then revert each row:

    H00 H01 H02 H03 H04 H05 H06
NR                          
1   33  28  98  97  0   0   0
2   29  24  22  98  97  0   0
3   78  76  98  97  0   0   0
4   16  15  98  97  0   0   0
5   81  72  70  98  97  0   0

This is the result I would like to get:

    H00 H01 H02 H03 H04 H05 H06
NR                          
1   97  98  28  33  0   0   0
2   97  98  22  24  29  0   0
3   97  98  76  78  0   0   0
4   97  98  15  16  0   0   0
5   97  98  70  72  81  0   0

I've tried varius shift and apply combinations without any success. Is there a simple way of achieving this?

Erfan · Answer 1 · 2020-10-06T16:54:08.843

We can use the justify function from this answer, then we use np.flip to reverse your data over the horizontal axis:

a = df.to_numpy()
a = np.flip(justify(a, side='right'), axis=1)
df = pd.DataFrame(a, columns=df.columns)

   H00  H01  H02  H03  H04  H05  H06
0   97   98   28   33    0    0    0
1   97   98   22   24   29    0    0
2   97   98   76   78    0    0    0
3   97   98   15   16    0    0    0
4   97   98   70   72   81    0    0

score 3 · Accepted Answer · answered Oct 06 '20 at 16:59

You can reverse the values that are greater than zero

def reverse_part(series):
  series[series > 0] = series[series > 0][::-1]
  return series

df.apply(reverse_part, axis=1, raw=True)

Out:

    H00  H01  H02  H03  H04  H05  H06
NR                                   
1    97   98   28   33    0    0    0
2    97   98   22   24   29    0    0
3    97   98   76   78    0    0    0
4    97   98   15   16    0    0    0
5    97   98   70   72   81    0    0

Shifting and reverting multiple rows in pandas dataframe

2 Answers2