The H-method is a data-based approach to obtain solution to mathematical models. In this site the H-mehod is presented together with different applications of the H-method. The H-method suggests that any mathematical solution, where the data are uncertain, should be build up in steps, where at each step the improvement in the solution is optimized with respect to the prediction aspect of the model. At each step the algorithms optimizes the improvement in the mathematical solution and the associated precision with the purpose of obtaining optimal predictions. The solutions obtained by the H-method are stable and provide with almost optimal prediction ability. The H-method is applicable to most traditional methods for data analysis, where data are uncertain.

Besides providing with better predictions than traditional methods, these methods provide with graphic procedures for analyzing different features in data. These graphic methods display the variation of the latent structure relative to the model, which is inherent in data. Using these graphic procedures we get thorough insight into the variation from each step of the numerical procedure used. They are available for any linear and non-linear modeling of data.

The H-method has been applied to many fields of applied sciences and shown its superiority to traditional methods in obtaining reliable predictions.

In industry the primary concern is related to the quality of prediction obtained. By focusing on the prediction aspect of the solution of the mathematical model in question, many successful applications have been developed.

Agnar Höskuldsson
Technical University of Denmark
ah@ipl.dtu.dk

This Web site is under development