Abstract
Using benchmark data, we investigated its accuracy using a spherical SOM．The following six data types were used: iris data, wine data, glass data, liver disease data, diabetes data, and breast cancer data. Each calculation was performed by normalizing the data to 0-1. For each method, each data was projected onto a spherical SOM, and the accuracy was judged by determining which group the boundary of each group of each data belonged to. As a result, the accuracy was highest for irises, wine, and breast cancer in that order. The rest had poor accuracy and many misclassifications.
