Causual Analysis of Data Using 2-Layerd Spherical Self-Organizing Map

Today, we are able to easily obtain data such as stock prices and market information through media such as the Internet. However, it is not so easy to come by useful information from this data. The reason for this is that there is a mix of many different kinds of information. Such a situation leads to difficulty in grasping trends in the data through just one rule, and so one must find a rule for each relevant factor. Therefore, the need for finding relevant factors is inevitable. For this reason, in our research we developed an algorithm that enables one to extract factors that have causal relationships with one another, and indicated its usefulness through experiments.