Techniques that reduce the number of features while preserving essential information. Methods like Principal Component Analysis (PCA) or t-SNE transform high-dimensional forex data (e.g. many technical indicators) into lower-dimensional form. Dimensionality reduction removes redundant or noisy inputs, speeding up training and helping generalization. For example, PCA might compress dozens of correlated currency indices into a few components. This improves model performance and interpretability by focusing on the most relevant signal features.

Back to Robots Glossary Index