We use a knowledge discovery approach to get insights over the features of the bin packing problem and its relationship in the performance of an evolutionary-based model of hyper- heuristics. The evolutionary model produces rules that combine the application of up to six different low-level heuristics during the solution of a given problem instance. Using the Principal Component Analysis (PCA) method, we visualize in two dimensions all instances characterized by a larger number of features. By over imposing features and hyperheuristic performance over the 2D graphs, it is possible to draw conclusions about the relation between the bin packing problem structure and the hyper-heuristics performance. Copyright is held by the author/owner(s).