No label defined (Q3689073)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3912258
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3912258 |
Statements
1984
0 references
group method of data handling
0 references
nonlinear identification
0 references
heuristics free
0 references
nonlinear modeling
0 references
Akaike's information criterion
0 references
optimal partial polynomials
0 references
optimal intermediate polynomials
0 references
0.8335021138191223
0 references
0.7943224906921387
0 references
[For the entire collection see Zbl 0559.00029.] Firstly we describe a revised GMDH (Group Method of Data Handling), named heuristics free GMDH of generating optimal partial polynomials in each selection layer, in which heuristics are not required such as dividing the available data into training data and checking data, predetermining the order of the partial polynomials, and predetermining the number of intermediate variables. This GMDH has many advantages compared with the conventional GMDH developed by \textit{A. G. Ivakhnenko} and \textit{Yu. V. Koppa} [Avtomatika 14, No.4, 69-78 (1969; Zbl 0197.440)], and has much better flexibility in constructing a final model. But the deficiency of this heuristics free GMDH is that a physically meaningful structure among the input and output variables cannot be identified. The reason for this is that, in this heuristics free GMDH as well as in the conventional GMDH, the partial polynomials are generated in each selection layer where the partial polynomials represent the relationship among the intermediate variables and the output variables, and that the polynomials as such are accumulated in the multilayered structure for constructing a final model. Secondly we describe a heuristics free GMDH of generating optimal intermediate polynomials in each selection layer instead of partial polynomials. The intermediate polynomials represent the direct relationship among the original input and output variables, and the polynomials as such are generated so as to minimize AIC (Akaike's Information Criterion) evaluated by using all the data. In this revised GMDH a physically meaningful model structure can be identified when the characteristics of the system are well imbedded in the data.
0 references