pandas group by multiple columns and remove rows based on multiple conditions

Question

I have a dataframe which is as follows:

imagename,locationName,brandname,x,y,w,h,xdiff,ydiff
95-20180407-215120-235505-00050.jpg,Shirt,SAMSUNG,0,490,177,82,0,0
95-20180407-215120-235505-00050.jpg,Shirt,SAMSUNG,1,491,182,78,1,1
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,3,450,94,45,2,-41
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,5,451,95,48,2,1
95-20180407-215120-235505-00050.jpg,DUGOUT,VIVO,167,319,36,38,162,-132
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,446,349,99,90,279,30
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,455,342,84,93,9,-7
95-20180407-215120-235505-00050.jpg,Shirt,GOIBIBO,559,212,70,106,104,-130

Its a csv dump. From this I want to group by imagename and brandname. Wherever the values in xdiff and ydiff is less than 10 then remove the second line.

For example, from the first two lines I want to delete the second line, similarly from lines 3 and 4 I want to delete line 4.

I could do this quickly in R using dplyr group by, lag and lead functions. However, I am not sure how to combine different functions in python to achieve this. This is what I have tried so far:

df[df.groupby(['imagename','brandname']).xdiff.transform() <= 10]

Not sure what function should I call within transform and how to include ydiff too.

The expected output is as follows:

imagename,locationName,brandname,x,y,w,h,xdiff,ydiff
95-20180407-215120-235505-00050.jpg,Shirt,SAMSUNG,0,490,177,82,0,0
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,3,450,94,45,2,-41
95-20180407-215120-235505-00050.jpg,DUGOUT,VIVO,167,319,36,38,162,-132
95-20180407-215120-235505-00050.jpg,Shirt,DHFL,446,349,99,90,279,30
95-20180407-215120-235505-00050.jpg,Shirt,GOIBIBO,559,212,70,106,104,-130

Answer 1

You can take individual groupby frames and apply the conditions through apply function

#df.groupby(['imagename','brandname'],group_keys=False).apply(lambda x: x.iloc[range(0,len(x),2)] if x['xdiff'].lt(10).any() else x)
df.groupby(['imagename','brandname'],group_keys=False).apply(lambda x: x.iloc[range(0,len(x),2)] if (x['xdiff'].lt(10).any() and x['ydiff'].lt(10).any()) else x)

Out:

    imagename   locationName    brandname   x   y   w   h   xdiff   ydiff
2   95-20180407-215120-235505-00050.jpg Shirt   DHFL    3   450 94  45  2   -41
5   95-20180407-215120-235505-00050.jpg Shirt   DHFL    446 349 99  90  279 30
7   95-20180407-215120-235505-00050.jpg Shirt   GOIBIBO 559 212 70  106 104 -130
0   95-20180407-215120-235505-00050.jpg Shirt   SAMSUNG 0   490 177 82  0   0
4   95-20180407-215120-235505-00050.jpg DUGOUT  VIVO    167 319 36  38  162 -132