CLASSIFICATION ALGORITHM C4.5 BASED ON PARTICLE SWARM OPTIMIZATION TO DETERMINE THE DELAY ORDER PRODUCTION PATTERN
Hendra setiawan M.Kom, Satria M.kom, Sabar Hanadwiputra M.kom, Kikim Mukiman M.Kom, Adi Suwarno M.kom
STMIK BANI SALEH BEKASI
Abstract
Production planning is an activity to establish a product to be produced. Planning aims to maximize service to consumers. PT Cedefindo is a unit of Martha Tilaar Group which is a manufacturing company in the field of cosmetic and herbal products in national and international scale. This company to achieve its success, is required to be able to carry out production planning well in order to increase trust to consumers. In 2015, PT Cedefindo experienced overdue of approximately 32 days in production planning. Planning that has been planned always outside the production schedule (leadtime) so that delivery of products to consumers is always too late. Therefore it is necessary to do late analysis so that it can know the delay of production order. Through the results of the analysis, the classification method for determining the pattern of production order delay is using two models, the classification algorithm C4.5 model and the Particle Swarm Optimization (PSO) classification algorithm C4.5 model. After the test with the two models obtained the results of the algorithm classification C4.5 to produce an accuracy of 80.17% and AUC value of 0.822 with the level of diagnosis of Good Classification, but after optimization with the algorithm based classification C4.5 Particle Swarm Optimization accuracy value Equal to 82,52% and AUC value equal to 0,855 with diagnosis level of Good Classification. So both methods have a difference of accuracy level of 2.35%.
Keywords: Leadtime, overdue, algorithm C4.5, Particle Swarm Optimization
Topic: Informatic and Information System