Year: 2000
Type: article
Abstract: Neural nets can be used for non-linear classification and regression models. They have a big advantage \nover conventional statistical tools in that it is not necessary to assume any mathematical form for the \nfunctional relationship between the variables. However, they also have a few associated problems chief of \nwhich are probably the risk of over-parametrization in the absence of P-values, the lack of appropriate \ndiagnostic tools and the difficulties associated with model interpretation. The first of these problems is \nparticularly important in the case of small data sets. These problems are investigated in the context of real \nmarket research data involving non-linear regression and discriminant analysis. In all cases we compare \nthe results of the non-linear neural net models with those of conventional linear statistical methods. Our \nconclusion is that the theory and software for neural networks has some way to go before the above \nproblems will be solved.
Author: Meyer, Denny
Institution: Massey University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7039646364