method.tex

\section{Data Analysis Method}
\label{sec:method}

We are interested in proposing an algorithm with a reduced and prioritized number of validation metrics based on previously acquired collection of validation data samples (i.e., our dataset). A common method to do this is to identify rules with disjunctive characteristics, in the form of decision trees, that lead to outcomes representative of the overall validation analysis. In our case, we define such outcome in terms of four \textit{validation categories} or \textit{classes} representative of the quality of the validation, namely: poor (P), fair (F), good (G), and excellent (E). The development of such decision trees requires a proper classification of the data, which can be done through a clustering process. This implies applying the \textit{labels} P, F, G and E to the data samples in our dataset. The following sections explain the data processing analysis we put in place to label the data samples in our dataset and subsequently obtain three alternative decision trees.

\input{clustering}

\input{decision-tree}