Determining erratic results of boric acid grain size with clustering analysis

Abstract

Control charts are commonly used to measure the variation occurring within the processes. In this study, Range (R) control chart was established in order to determine whether the results from the sieve analysis conducted in consecutive days in boric acid manufacture plant were under control or not. When the R control chart was examined, it was observed that some points were incompatible with other points. It is significant for the process to determine whether these incompatible points are erratic results or not. Therefore in this study, single linkage method, one of Agglomerative hierarchical clustering methods, was used to determine whether the incompatible points identified in improper locations were erratic results or not. As a result of this method, erratic results were determined and eliminated from the process. After R control chart was drawn again and it was determined that the re-drawn R control chart displayed the process characteristic. It was suggested in conclusion that it was a necessity to establish R control charts only after determining erratic results and eliminating them from the process by using single linkage method.