Apache Spark Deep Learning Cookbook

There's more...

While ytrain_array and ytest_array are both labels in a matrix format, they are not ideally encoded for deep learning. Since this is technically a classification model that we are building we need to encode our labels in a manner for them to be understood by the model. This means that our ratings of 0 through 5 should be encoded as 0 or 1 values, based on their value elements. Therefore, if a rating received the highest value of 5, it should be encoded as [0,0,0,0,0,1]. The first position is reserved for 0, and the sixth position is reserved for 1, indicating a value of 5. We can make this conversion using keras.utils and convert our categorical variables to one-hot encoded variables. In doing this, the shape of our training ...