Predicting customer conversion with deep learning algorithms

Deep learning has become a hot topic and its popularity and usage are rising, as deep learning models are proven to work well when data have complex relationships within the variables and learn or extract features autonomously from the data, even though tree-based models are also very frequently used and powerful for predictive modeling. We touched on deep learning in Chapter 5 when we used pre-trained language models for sentiment analysis and classification. In this section, we are going to build on that knowledge and experiment with developing deep learning models for predictive modeling and, more specifically, for making predictions on which customers are likely to convert.

Deep learning is basically an artificial neural network (ANN) model with lots of hidden and complex layers of neurons, or, in other words, a deep ANN. An ANN is a model inspired by the biological neural networks in animal and human brains. An...