DAERAH KEPERCAYAAN UNTUK LINTASAN RIDGE RESPON GANDA

INAYATUL KHOIRIYAH, Danang Teguh Qoyyimi, M.Sc., Ph.D.

2019 | Skripsi | S1 STATISTIKA

Analisis ridge mempunyai peranan penting dalam metode permukaan respon, yang mana dapat mengeksplorasi kondisi operasi optimal dari faktor-faktor percobaan. Model seemingly unrelated regression (SUR) diaplikasikan untuk mendapatkan model permukaan respon ganda. Model SUR dapat menentukan model sesuai untuk tiap respon sebagai suatu sistem persamaan yang mempunyai korelasi satu sama lain. Respon-respon hasil estimasi ditrasnformasi melalui fungsi desirability yang kemudian dioptimasi dengan kendala-kendala tertentu. Lintasan ridge dihasilkan dari plot antara desirability maksimum dan radius, yaitu nilai maksimum overall desirability dan jarak faktor-faktor optimal dari pusat daerah percobaan. Plot berupa lintasan ridge tersebut beserta interval kepercayaan disebut sebagai daerah kepercayaan untuk lintasan ridge repon ganda.

Ridge analysis has an important role in the response surface methodology, which can explore the optimal operating conditions of the experimental factors. The seemingly unrelated regression (SUR) model was applied to fit a multi-response surface model. The SUR model could fit the model of each response as a system of equations that have correlation with each other. The estimated responses are transformed via desirability function, then optimized with certain constraints. The ridge path is defined as a plot between the maximum desirability and radius, that are the maximum value of overall desirability and the distance of the optimal factors from the center of the experiment region. The plot of the ridge path along with the confidence interval is called as the confidence region for a multi-response ridge path.

Kata Kunci : multi-response surface methodology, seemingly unrelated regression, desirability function, generalized reducent gradient.