Dimension Mismatch Error during dot product in Python

Question

Dimension Mismatch Error during dot product in Python

Ambar Walia

2022年5月31日 09:54

I have two matrices user_vecs and item_vecs

I am trying to take the dot product of the two to build a recommendation engine:

The shape of the two vectors are as follows:

user_vecs.shape
(20051, 20)

item_vecs.shape
(20,1808)

When I take the dot product of the transpose as follows:

a = user_vecs.dot(item_vecs.transpose())

I get the following error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
ipython-input-41-f4cd01978711 in module
---- 1 a = user_vecs.dot(item_vecs.transpose())

C:\ProgramData\Anaconda3\lib\site-packages\scipy\sparse\base.py in dot(self, other)
    362 
    363         
-- 364         return self * other
    365 
    366     def power(self, n, dtype=None):

C:\ProgramData\Anaconda3\lib\site-packages\scipy\sparse\base.py in __mul__(self, other)
    479         if issparse(other):
    480             if self.shape[1] != other.shape[0]:
-- 481                 raise ValueError('dimension mismatch')
    482             return self._mul_sparse_matrix(other)
    483 

ValueError: dimension mismatch

I understand that the dimensions of the two matrices are not matching, but the transpose should have fixed that. Why am I still getting this error?

Topic python machine-learning

Category Data Science

haddagart · Accepted Answer · 2022年5月31日 09:54

Mathematically speaking this is matrix dot product. i.e. the number of columns in user_vecs must match the number of rows (lines) in item_vecs. In this case the matching is already established since performing a dot product on (20051, 20) by (20,1808) is mathematically feasible. You need to just to put on the product directly.

Transpose is the source of the error in the code above

a = user_vecs.dot(item_vecs)

Vamshi Kumar Kurva · Accepted Answer · 2021年8月24日 00:53

Firstly user_vecs and item_vecs are matrices, not vectors, since both the shapes are of two dimensions. It seems like you are trying to multiply both the matrices, so you can simply use user_vecs.dot(item_vecs). No need to transpose item_vecs. You can also use np.matmul(user_vecs, item_vecs). Refer to this thread for more details.

Dimension Mismatch Error during dot product in Python

About