Comparative Analysis Of Different Data Analysis Methods In Ancient Ceramics Provenance And Chronology Research

Our country has a long history and splendid civilization,Ancient porcelain is animportant part of the silicates cultural relies.Its invention, production, currency andapplication have an important position in the research of the history of humancivilization,especially the history of early civilization.People in the process of scienceand technology of ancient ceramics research and identification, using a variety ofanalysis methods measured the date of elemental composition of the body and glazeof ancient ceramics,which is the first step of in working.How to really put in a lot ofseemingly haphazard hidden digital information centralized, extraction, refining it, itis very important to study and identification of ancient ceramics[1,2].Due to a widerange of ancient ceramic products,its origin and the development porcelaintechnology and art are also different,the chemical composition of ancient ceramics inour country is very complex,so data variables need to be screened and components toidentify pretreatment.Maximize the induction information that we need from the dataanalysis diagram is a very important in identification and analysis process[2].However,the data of the ancient ceramic samples has its own characteristics,not only thediversity of their ancient ceramics,but also effected by ancient ceramics elementscomposition analysis technology,in addition,different methods of data analysis alsohas its own characteristics and scope,so according to the date of ancient ceramicsamples’ own characteristics and specific research purposes,how to effectively selectand apply the appropriate data analysis methods to carry out the analysis of ancientceramics date,which become a hot issue of ancient ceramic technology researchers areconcerned.The paper intents to ancient ceramics provenance and chronology,origincharacteristics and evolution of ancient ceramics,because Longquan celadon not onlyhave been loved widely by domestic and international ceramic lovers,but also imitatedby the kiln workers from different places,Among all of the imitated celadon,the mostrepresentative products appeared in Ming and Qing dynasties,and both are close inappearance,traditional identification method is difficult to distinguish.So usingchemical composition of body and glaze elements of Jingdezhen imitated Longquanand Longquan celadon for the study.Choosing multivariate statistical discriminantanalysis, artificial neural networks and support vector machine several differentmethods of data analysis for its data analysis,obtains the classification characteristicsof chemical composition of body and glaze of samples,establishments thediscriminant formula of various types of samples,to a certain extent,confirmed the scope and characteristics of application of different methods of data analysis, providessome reference of the application of a variety of different data processing methods inancient ceramics provenance and chronology.Moreover,in this paper artificial neuralnetwork and support vector machine to be compared with multivariate statisticaldiscriminant analysis,discusses the differences and applicability,easier for people toaccording to the characteristics and condition of data,choosing more reliability andaccuracy of methods to achieve the purpose of data analysis.This provides somereferences for improving the rational utilization of different data analysis methods onancient ceramic technology research.