Title: Predicting symbolic interval-valued data through symmetrical nonlinear regression
Authors: Dailys Maite Aliaga Reyes; Renata Maria Cardoso Rodrigues De Souza; Francisco José De Azevêdo Cysneiros
Addresses: Centro de Informática, Universidade Federal de Pernambuco, Recife, PE, Brazil ' Centro de Informática, Universidade Federal de Pernambuco, Av. Jornalista Aníbal Fernandes s/n, 50740-560, Recife, PE, Brazil ' Departamento de Estatística, Universidade Federal de Pernambuco, Av. Professor Luiz Freire s/n, 50740-540, Curado, Recife, PE, Brazil
Abstract: We proposed a symmetrical nonlinear regression model to fit interval-valued data. An important feature of this new model is that the estimate and prediction are less sensitive in the presence of outliers than a nonlinear model proposed in the literature. Monte Carlo simulation studies have been developed to investigate the performance of the model on different scenarios in precense of some percentage of outliers. The results based on the mean magnitude of the relative errors are presented and discussed. The model was fitted to one real symbolic dataset with noticeable interval outliers, and the forecast accuracy has been considered.
Keywords: symmetrical nonlinear regression; interval-valued data; symbolic data analysis; outliers.
DOI: 10.1504/IJBIDM.2017.084283
International Journal of Business Intelligence and Data Mining, 2017 Vol.12 No.2, pp.175 - 189
Received: 04 Jan 2017
Accepted: 03 Feb 2017
Published online: 23 May 2017 *