| dc.description.abstract | This study aimed to design an R package that implements the demerit chart based on fuzzy clustering. The R package called dissertation1 was created using R with functions demerchart. For the function demerchart to work, it depended on other packages for the creation of fuzzy cmeans, dataset and adding of a dataset to the package. The dependencies are devtools, ppclust, factoextra, ggplot2, dplyr, cluster, psych, qcc, rlang, fclust and inaparc. The R dataset used was “orange juice” dataset with 54 samples each of sample size 50, number of defects per sample and trial.
The Package dissertation1 returns a summary of results from the function demerchart 54 objects each of size 50 and of which 4 clusters were selected. The 4 initial clusters selected randomly were returned and the final cluster prototypes after the fuzzy clustering algorithm were run. It also shows the first four top and four bottom of membership values that are used together with the number of defects and weights per class to compute the control limits used in constructing the demerit control chart. The demerchart function by default rounds off the LCL to zero if negative results are returned.
Therefore, the package dissertation1 returns a demerit control chart based on fuzzy c-means clustering that is more sensitive to assignable causes. | en_US |
