X . Self Organizing Maps , a Visual Exploration Tool in Datamining

We humans are good at detecting visual patterns. Driving a car, scanning the headlines of newspapers or noticing that a single tile is out of alignment in your bathroom, during most of our daily activities we rely heavily upon the information we can extract from what our eyes can see. Our affinity for visual imagery can be used to effectively and efficiently convey information. This has been done for ages. Graphs and charts have been used to give better insight into data than is possible with numbers alone. This paper is about Self-Organizing Maps (SOMs), an algorithm that can be used to turn data into something we can see and understand: a twodimensional map. Tools like SOMs are invaluable in datamining. They provide the means to make high-dimensional data understandable and, perhaps more importantly, explorable for the user. These visual exploration tools give the opportunity to first get a look and feel of the complex data; they do not require the user to ask the right questions right away. First of all, this paper will elaborate on what SOMs are and how they work, followed by some examples of SOMs applied in datamining and closing off with a look at the future of visual exploration tools.