search query: @indexterm Neural networks / total: 121
reference: 61 / 121
Author: | Heimel, J. P. |
Title: | Konnexionistische Kaufakt- und Markenwahlmodelle. |
Journal: | Schmalenbachs Zeitschrift für Betriebswirtschaftliche Forschung
1998 : VOL. 50:7/8, p. 596-613 |
Index terms: | MARKETING MODELS ECONOMETRIC MODELS NEURAL NETWORKS FORECASTING CONSUMER BEHAVIOUR |
Language: | ger |
Abstract: | Artificial neural networks can be seen as nonlinear generalizations of conventional statistical or econometric models. This paper studies goodness-of-fit and forecasting performance of neural networks with one hidden layer. The central dependent variables regarded are purchase incidence and brand choice within the context of consumer non- durables. The connectionist models are compared to the well- known NBD and conditional logit models. Models are estimated on the basis of household panel data. On the whole, the neural models achieve better results than their conventional counterparts. Influence of predictors is interpreted using a distribution of elasticities or change rates. |
SCIMA