Performance of Heuristics for Classifying Leftovers from Cutting Stock Problem

The one-dimensional cutting stock problem is defined in the literature as a branch of the classic cutting stock optimization problem, involving one dimension in the cutting process, like cutting bars. The bar cutting optimization problem can generate leftovers - reusable- or losses - disposable. The objective of this paper is to compare the performance of OptimizationDistBSP and OptimizationTREE heuristics (proposed in [2]) for classifying leftovers or losses, from the cutting stock problem (specifically from cutting one-dimensional bars), using the dataset proposed in [3], since this dataset allows the application of Machine Learning methods, which are: Logistic Regression, Naive Bayes, Decision Tree and Random Forest, to classify the output data as leftover or loss. Results show that the OptimizationDistBSP and OptimizationTREE heuristics provide better performance in the classification task than the Greedy heuristic used in [2]. Thus, we can conclude that the heuristics can be applied in a more realistic problem, using bars of different sizes, and the dataset can be validated, providing good results for the classification using heuristics other than Greedy.