C3_W2_Collaborative_RecSys_Assignment data Order issue

Course Q&A Machine Learning Specialization Unsupervised Learning, Recommenders, Reinforcement
1

Hello,

I have a question about this line in the code: ix = tf.argsort(my_predictions, direction=‘DESCENDING’)

image
image1203×608 41.3 KB

Not sure how the function sorted the indices in the m_predictions array. It seemed that for 5 star ratings, it gave random order to the indices. I would have expected the order to be like this: [2700,1150,929,793,622,366,246]

image
image1183×431 36.6 KB

I tried the function on a smaller array but it seems that for similar values in the array, it sorts them in an ascending order towards the end of the result array. I am a little confused on how tf.argsort(my_predictions, direction=‘DESCENDING’) work.

image
image1165×156 8.44 KB

Thanks

2

Hello @Mo_Okasha,

There is a difference between tf.sort and tf.argsort. The former returns the sorted values, whereas the latter returns the indices of the sorted values. Please experiment both on the same array.

Raymond

3

Sorry, you can ignore this. I was confusing the data in my_ratings with my_predictions. Thanks.