Comparison of two wood plastic composite extruders using bootstrap confidence intervals on measurements of sample failure data

摘要

Purpose: To compare the modulus of elasticity (MOE) and the modulus of rupture (MOR) of the wood plastic composite board from two production lines using the confidence intervals of the percentiles by the application of bootstrapping. Summary: Bootstrapping is a computer-intensive re-sampling method used to estimate statistical properties which are difficult to estimate analytically and the method is free from parametric assumptions however can be used when parametric assumptions are necessary. There may be situations when parametric distributions are known, it may be difficult to estimate certain statistic. Bootstrapping can be used in such situations. There are number of bootstrapping methods both parametric and non-parametric are available. The most commonly used method is to collect number of samples with replacement of the same size as that of the original sample. The statistic of interest is calculated from the subsamples and the distribution of the statistic is approximately found using histogram which is the bootstrap distribution. The parameters of interest can be calculated using this distribution. This paper illustrates the bootstrapping method, both parametric and non-parametric to derive the confidence intervals of many percentiles such as first, fifth and tenth which form important quality indices. The method is used to compare the perpendicular pressure required to permanently deform a wood plastic composite board (MOE) and the perpendicular pressure required to rupture the board (MOR). Samples were collected from two production lines to obtain MOE and MOR and bootstrapping is used to compare the results.