Optimization of Atmospheric Plasma Surface Modification Process Using Decision Trees

Abstract

1 Textile Material and Processes Research Unit, University of Monastir, Tunisia \u00a0 (Received October 11 2019, accepted November 28 2019) Decisions trees are one of the most commonly used data mining techniques to practically solve classification and prediction problems. They have tree shaped structures in which construction of trees is simple and unlike the logistic regression models, decision tree results can be easily understood by the users. In this study, a \u00a0decision tree induction algorithm known as CART (Classification and Regression Trees) has been employed in order to better understand the influence of plasma parameters adjustment on polypropylene (PP) film\u2019s hydrophilic surface properties. The cross-validation method was used for pruning the decision tree. The root mean square errors (RMSE) and correlation coefficients (R) for training and test subsets were used in order \u00a0to \u00a0get \u00a0the \u00a0best \u00a0fitting model. \u00a0The \u00a0obtained \u00a0decision \u00a0tree \u00a0regression \u00a0model \u00a0showed \u00a0excellent \u00a0learning performance and achieved good predictive accuracy. \u00a0