Dataset for machine learning based approximations of strip packing / nesting heights

Published: 13-10-2020| Version 1 | DOI: 10.17632/scr79cbxxd.1
Christian Gahm


Solving strip packing or nesting problems with highly irregular shapes is a complex and computation time intensive task. In some cases, it is adequate to estimate the strip height instead of solving the nesting problem completely. The file Nesting sample data.csv/*.xlsx consists of 88,200 (complex) nesting instances generated and used in the paper "Approximate anticipation of base-level reactions by machine learning techniques used to substitute the solving of complex nesting problems". The first 14 columns describe details of the nesting instances (e.g., item types used in the instance, the width of the object, used in which phase of the machine learning process etc.), whereat the column "Nesting Solution" is the computed height of the instance. All the remaining columns are numerical features which can be used as a input in the machine learning process (for details see the reference paper). Besides the two data files, a jupyter notebook is provided to illustrate the usage of the data.