Applying an integrated SOM model on studying corporate governance data

Corporate governance has gradually become an important consideration in investment decisions of stock market and credit evaluation of financial market since several notorious business scandals, such as Enron's inside information trading in the US and executive entrenchment of Rebar conglomerate in Taiwan. However, corporate governance is presented by a large amount of data, which involves the following facets: managerial ownership, external ownership, board size, board composition, financial transparency, executive compensation, etc. To utilize these data, this study applied an integrated self-organizing map (SOM) model, which has been successfully used for visualizing and clustering high-dimension data in several applications, to study corporate governance of Taiwan's semiconductor companies by recognizing implicit patterns of these data. The research results demonstrate that eight types of corporate governance have been identified by the SOM model, and features selected by LabelSOM are used to interpret the eight styles of companies.