EcoSta 2023: Start Registration
View Submission - EcoSta2023
A0206
Title: The order-of-addition experiments on the adjacency relationship Authors:  Jian-Feng Yang - Nankai University (China) [presenting]
Abstract: With the development of science and technology, the scale of data has become huge. The design of experiments plays a more and more critical role in different areas. In many cases, the addition order of components, which is the order-of-addition (OofA) problem, is studied. The adjacency relationship (AR) between components usually has an important impact on the final results. However, there is still very little research on this part, while most of the existing order-of-addition experiments focus on the relative or absolute positions between components. How AR between components affects the response is considered. Starting from the Traveling Salesman Problem, the aim is to propose an AR model and presents an algorithm for inferring the optimal order. In addition, a loop-full design, which is different from the OofA full design, is proposed. The properties of the loop-full design and the D-optimality under the AR model are given. A kind of minimal-point design is also put forward, which greatly reduces the run size and performs well in designs with the same run size. Based on the minimal-point design, the study brings up an improved algorithm to improve D-efficiency. The simulation results show the effectiveness of the model and design.