# Group Method of Data Handling **Domain:** Machine Learning / Systems Identification / Cybernetics **Doc Type:** Technical Method Node **Maturity:** Historical ## Definition The **Group Method of Data Handling**, or **GMDH**, is a family of self-organizing modeling procedures developed by [[wiki/Alexey Ivakhnenko|Alexey Ivakhnenko]]. Candidate combinations of input variables are generated in layers, fitted to data and selected according to external performance criteria. ## Historical Significance GMDH joined a learned multilayer architecture with a separate testing set and structural selection. Ivakhnenko described this as heuristic self-organization and decision regularization. ## See Also [[wiki/Held-Out Validation|Held-Out Validation]], [[wiki/Overfitting|Overfitting]], [[wiki/Artificial Neural Networks|Artificial Neural Networks]]