Proposing a new approach to the selection of material portfolio using a combination of data mining and optimisation methods

The present paper aims to provide a new framework for the selection of a portfolio of materials. This paper shows that, compared with the traditional methods in the selection of materials, how the new materials are analysed based on new ideas. The case study is materials required for production of tile glaze in both traditional and modern methods. Modelling in this study was done based on mathematical description approach. The results of execution of feature selection algorithm revealed that important factors in the selection of glaze for production of tile in both traditional and modern methods include cracking, self-cleaning, uniformity, water absorption, and market potential. In addition, the results of K-means algorithm showed that all the materials of choice for production of tile glaze are not placed in a single cluster. Therefore, each cluster should be evaluated separately. Unlike the classical approaches to the selection of materials, in the new approach, candidates for the selection of tile glaze are firstly clustered by K-means algorithm. Each cluster is independently ranked using free disposal hull model. Free disposal hull is a mathematical programming model based on data envelopment analysis. The final optimised portfolio of materials was determined using the genetic algorithm.