Multiple Linear Regression Model on Compression Strength of Corrugated Boxes

摘要

Using a multiple linear regression model, we analyze test results based on a sample of thousands of batches of corrugated boxes to establish the relation between compression strength and a series of variables such as bursting strength, edgewise crush strength, puncture resistance and ply adhesive strength. Based on the non-linear regression model results, we reveal that the relation between compression strength and other independent variables is more akin to that of a power function. We undertake a logarithmic transformation on the best fitting non-linear regression model to establish a formula for predicting compression strength. To test how well our formula predicts compression strength, we compare real test results with values predicted both by our formula and by traditional formulas such as those of Kellicutt, Makee and Wolf. We find the values predicted by our model are closer to those derived in real tests than are those predicted by the other three formulas considered.