haku: @indexterm neural networks / yhteensä: 121
viite: 35 / 121
| Tekijä: | Gierl, H. Schwanenberg, S. |
| Otsikko: | Clusteranalyse mittels SOFM |
| Lehti: | Marketing: Zeitschrift für Forschung und Praxis
2001 : VOL. 23:2, p. 129-141 |
| Asiasana: | Neural networks |
| Kieli: | ger |
| Tiivistelmä: | There are two concepts of neural networks which became very popular: multi layer perceptrons (MLP) and self organizing feature maps (SOFM). The advantages of the MLP for segmentation purposes are already analysed by Hruschka/Natter in detail. In this article we evaluate the usefulness of the second approach (SOFM) for cluster analysis. Based upon the results of a Monte-Carlo simulation we show that the Ward algorithm should be preferred to identify the number of clusters. But the application of a SOFM seems to be superior to k-means if observations are assigned to clusters. |
SCIMA