Decision Trees as a Tool for Data Analysis. Elections in Barcelona: A Case Study

Decision trees are inductive learning methods that construct a domain model easy to understand from domain experts. For this reason, we claim that the description of a given data set using decision trees is an easy way to both discover patterns and compare the classes that form the domain at hand. It is also an easy way to compare different models of the same domain. In the current paper, we have used decision trees to analyze the vote of the Barcelona citizens in several electoral convocations. Thus, the comparison of the models we have obtained has let us know that the percentage of people with a university degree is the most important aspect to separe the neighbourhoods of Barcelona according to the most voted party in a neighbourhood. We also show that in some neighbourhoods has always won the same party independently of the kind of convocation (local or general).