Group method of data handling
The Group Method of Data Handling is a statistical method to find an optimal specification or model for a data set. It was developped by the Ukranian mathematician Alexey Grigorevich Ivakhnenko in the 1970s. It is a general data mining method. It can also be called polynomial neural network or inductive/statistical learning.
The GMDH algorithm optimises by sorting-out of gradually complicated models and selecting the best solution by finding the minimum of the external criterion characteristic.