In the following command, we can see how the previous distribution affects the subscription:

%sql select previous, count(1), y from trainData group by previous,y order by previous,y

Like the previous chart, this one confirms that most of the clients were not previously contacted for this campaign.