Hey there, did you hear the discussion between Mr. Clough and Mr Sheene? And who do you think was the next person Mr. Sheene talked to after Mr. Clough? Yeah, you are guessing right, it was me: Andy, I want the list on my desk in two hours. Mr. Sheene was actually quite upset by Mr. Clough suggesting that one of us spread the word about the analyses. 

That said, what we have to do now? Well, first of all, we still have to fit random forest on our data, in order to complete our data modelling strategy. Finally, we will employ all of our estimated valid models on the full list of customers pertaining to the Middle East area. 

The result of this application will be the list of customers enriched with our model prediction.

What? How are we going to merge predictions from our different models? We are going to leverage ensemble learning techniques for that. But let's keep things in their order—we still have to fit two more models, and time is running out. 

Time to hurry up now and fit our last model: random forest. As usual, I will first of all show you the intuition behind it, then go into the math, and finally show you how to apply it. But before doing this, we need to start from the basic building block of any random forest: decision trees.