We looked at what is a collaborative filtering method and went into different collaborative filtering strategies. We introduced the R package recommenderlab to perform collaborative filtering. We leveraged the Jester5k dataset to demonstrate a collaborative filtering algorithm. We looked at the random model, popular model, item-based similarity, user-based similarity models, and factor models. We introduced the concept of evaluating the performance of a recommender system before deploying it. We demonstrated the steps to split the datasets in order to evaluate our model performance.
In the next chapter, we will be looking to build Deep Neural networks using the MXNet package for time series data. We will introduce the package MXNet R and proceed to build a deep neural network.