如何根据以下代码创建相似性矩阵？

Question

I'm trying to use the gower function from this link https://sourceforge.net/projects/gower-distance-4python/files/ . I'm trying to apply it to my dataframe of categorical variables. However I can see that when i use the gower_distances function i have some non-zero values in my diagonals ( i need them to all be 0).

I've been trying to de-bug the code. I think i know where this is happening and it's occuring in the _gower_distance_row function. There is this line of code which i don;t understand sij_cat = np.where(xi_cat == xj_cat,np.zeros_like(xi_cat),np.ones_like(xi_cat)). But i will present it in a easier format to understand.

Say i have:

xi=np.array(['cat','dog','monkey'])
xj=np.array([['cat','dog','monkey'],['horse','dog','hairy']])
sij_cat = np.where(xi == xj,np.zeros_like(xi),np.ones_like(xi))

I get this as my result:

array([['', '', ''],
       ['1', '', '1']], dtype='<U6') 

since i am comparing cat with cat i want to assign zero, and where it is different eg cat vs horse and monkey vs hairy it should be 1. I don't get why in the above result i am getting ''? i want zeroes here. How do i fix this?

Answer 1

np.logical_not(xi == xj).astype(int)

output will be:

array([[0, 0, 0],
       [1, 0, 1]])

explanation: np.logical_not changes True to False and False to True and astype(int) changes to 0 and 1