The matrix factorization model for Retailrocket recommendations

Now let's create a matrix factorization model in Keras:

1. Store the number of visitors and items in a variable, as follows:

n_visitors = events.visitorid.nunique()
n_items = events.itemid.nunique()

2. Set the number of latent factors for embedding to 5. You may want to try different values to see the impact on the model training:

n_latent_factors = 5

3. Import the Input, Embedding, and Flatten layers from the Keras library:

from tensorflow.keras.layers import Input, Embedding, Flatten

4. Start with the items – create an input layer for them as follows:

item_input = Input(shape=[1],name='Items')

5. Create an Embedding representation layer and then flatten the Embedding layer to get the output in the number of latent dimensions that we set earlier:

item_embed = Embedding(n_items + 1,
                           n_latent_factors, 
                           name='ItemsEmbedding')(item_input)
item_vec = Flatten(name='ItemsFlatten')(item_embed)

6. Similarly...