Data Streams for Unsupervised Analysis of Company Data

Financial data is increasingly made available in high quantities and in high quality for companies that trade in the stock market. However, such data is generally made available comprising many distinct financial indicators and most of these indicators are highly correlated and non-stationary. Computational tools for visualizing the huge diversity of available financial information, especially when it comes to financial indicators, are needed for micro and macro-economic financial analysis and forecasting. In this work we will present an automatic tool that can be a valuable assistant on this process: the Ubiquitous Self-Organizing Map (UbiSOM). The UbiSOM can be used for performing advance exploratory data analysis in company fundamental data and help to uncover new and emergent correlations in companies with similar company financial fundamentals that would remain undetected otherwise. Our results show that the generated SOM are stable enough to function as conceptual maps, that can accurately describe and adapt to the highly volatile financial data stream, even in the presence of financial shocks. Moreover, the SOM is presented as a valuable tool capable of describing different technological companies during the period of 2003–2018, based solely on four key fundamental indicators.