Matrix factorization basics in ML Python - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a matrix factorization model using numpy.

ML Python

import numpy as np

# Initialize matrices
R = np.array([[5, 3, 0], [4, 0, 0], [1, 1, 0]])
num_users, num_items = R.shape
K = 2
P = np.random.rand(num_users, [1])
Q = np.random.rand(num_items, K)

Drag options to blanks, or click blank then click option'

A1

Bnum_users

Cnum_items

DK

Attempts:

3 left

💡 Hint

Common Mistakes

Using num_users or num_items as the second dimension of P instead of K.

Setting the dimension to 1 which is incorrect for latent features.

The matrix P should have dimensions (num_users x K), where K is the number of latent features.

2fill in blank

medium

Complete the code to compute the predicted rating matrix by multiplying P and Q transpose.

ML Python

predicted_ratings = np.dot(P, [1])

Drag options to blanks, or click blank then click option'

AQ

BR

CQ.T

DP.T

Attempts:

3 left

💡 Hint

Common Mistakes

Multiplying P by Q without transpose causing dimension mismatch.

Using P.T or R which are incorrect for prediction.

To get the predicted ratings matrix, multiply P (users x features) by Q transpose (features x items).

3fill in blank

hard

Fix the error in the gradient descent update step for matrix factorization.

ML Python

for i in range(num_users):
    for j in range(num_items):
        if R[i][j] > 0:
            eij = R[i][j] - np.dot(P[i, :], Q[j, :].T)
            for k in range(K):
                P[i][k] += [1] * (2 * eij * Q[j][k] - 0.01 * P[i][k])

Drag options to blanks, or click blank then click option'

A0.002

B0.02

C-0.002

D-0.02

Attempts:

3 left

💡 Hint

Common Mistakes

Using a negative learning rate causing the model to diverge.

Using too large a learning rate causing unstable updates.

The learning rate should be positive to move in the direction that reduces error.

4fill in blank

hard

Fill both blanks to complete the error calculation and regularization term in matrix factorization.

ML Python

error = 0
for i in range(num_users):
    for j in range(num_items):
        if R[i][j] > 0:
            error += (R[i][j] - np.dot(P[i, :], Q[j, :].T))[1] 2
            for k in range(K):
                error += 0.01 / 2 * (P[i][k][2] 2 + Q[j][k][2] 2)

Drag options to blanks, or click blank then click option'

A**

B*

C+

D-

Attempts:

3 left

💡 Hint

Common Mistakes

Using '*' instead of '**' which multiplies but does not square.

Using '+' or '-' which changes the meaning of the formula.

The squared error uses the power operator '**2' to square the difference and the regularization terms.

5fill in blank

hard

Fill all three blanks to create a dictionary of predicted ratings for items with predicted rating above 3.

ML Python

predicted_dict = {item: predicted_ratings[user][[1]] for user in range(num_users) for item in range(num_items) if predicted_ratings[user][item] [2] 3 and predicted_ratings[user][item] [3] 5}

Drag options to blanks, or click blank then click option'

Aitem

B>

C<

Duser

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'user' instead of 'item' as dictionary keys.

Using wrong comparison operators causing incorrect filtering.

The dictionary keys are item indices, and the condition filters ratings greater than 3 and less than 5.