Summary
In this chapter, we prepared key information to create the input dialog of a chatbot. Using the weights of an RBM as features constituted the first step. We saw that we could use neural networks to extract features from datasets and represent them using the optimized weights.
Processing the likes/dislikes of a movie viewer reveals the features of the movies that, in turn, provide a mental representation of a marketing segment.
PCA chained to an RBM will generate a vector space that can be viewed in TensorBoard Embedding Projector in a few clicks.
Once TensorBoard was set up, we analyzed the statistics to understand the marketing segment the dataset originated from. By listing the points per feature, we found the main features that drove this marketing segment.
Having discovered some of the key features of the marketing segment we were analyzing, we can now move on to the next chapter and start building a chatbot for the viewers. At the same time,...
