Self-Organizing Maps for Visualizing and Clustering Financial Features
Summary
The article introduces self-organizing maps (SOMs), an unsupervised neural-network method that arranges high-dimensional input vectors on a two-dimensional grid. Each node has a weight vector; training repeatedly finds the best-matching node for a sample and adjusts that node and nearby nodes toward the sample. The neighborhood radius and learning rate decline during training, producing a map whose nearby regions represent similar inputs.
Examples in MetaTrader 5 include clustering RGB colors, displaying alternative grid and cell shapes, and applying SOMs to financial data. The article describes using the map to group similar characteristics and support correlation analysis, and outlines reading CSV data and rendering map images. These examples illustrate implementation and visualization rather than establish a predictive trading edge. Map quality depends on choices such as training data, grid size, and training settings, and the article cautions that drawing intermediate bitmap images can be slow.
Key ideas
- A self-organizing map learns without labeled target outputs.
- Training identifies a best-matching node and adjusts its weights and those of nearby nodes.
- The resulting grid visualizes similarity in data with many dimensions.
- Examples cover color clustering and feature maps for financial data in MetaTrader 5.
- The article demonstrates implementation and visualization, not evidence of profitable trading.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.