I tried to imitate the PySpark ALS Code from this Kaggle https://www.kaggle.com/vikashrajluhaniwal/matrix-factorization-recommendation-using-pyspark
I have noticed that when you use the code
model.recommendForAllUsers(3).show()
the output only consists of the no. of users in the training set (58971 users).
I was wondering how do you also get the recommendation of the test set (other 44819 users.) ?? I've tried searching for other tutorials but I still do not know how to get the prediction of the full dataset.
Or do I just use the parameter of the best model and just train the whole dataset again without splitting ?
Thank you so much for your answers.