]Analysis: 1. level
COCO-OPTI1 online, additive - standard version
Preconditions for using COCO-OPTI1:
- There is a dual OAM,
- the upper half of it consists of the primary (positive and integer) values of the Y, and the ranks (X) connected to the primary observations (based on 'the more is the better' principle),
- while the lower (inverse) half is ranked using the 'the less is the better' principle, where the values of the Y are equal with the upper half's Y-values, and the ranking (for X) is continuous both in the direct and the inverse view.
- The Y do not have to be zero, even if any of the attributes are missing (=additive effect).
- It is not necessary/possible to handle the surcharges and the attribute proportions.
- Columns that have the same ranking vector were reduced in order to speed up the execution (cf. repetitive columns lose their effect during the execution).
- Ceteris paribus functions may be optional (cf. polynomials).
- The first rows of stairs do necessarily mean the model's (genetic) potential.
- It is not necessary to give the inverse form in the case of each attribute.
- The execution of MY-X FREE program consists of 4 parts: the definition of the curves of those direct and indirect effects that designate the direct and inverse proportions, and the definition of those curves that are formed around the priorities of the direct (\) and indirect (/) effects, and finally the aggregation of the mutually negative parts of the two mechanisms that draw a partial relation. The estimation error balanced collection of these constitute the optimizing model itself, that can have a significant effect to the value of square error, and the correlation between the estimations and the facts, and the ceteris paribus curves: (demo)
- The 4 partial model can be united at request, keeping the hybridization effects in mind in concern of the correlation/error...
Applications:
- Building models that allows presuming optimum-effects (e.g. production functions).
- Simulation of decision trees (viz. searching for polynomial-parts that can be explained with the IF/THEN principle).
- Revelation of system-behavior.
- Making forecasts.
- Not advised: in case of benchmarking and price/performance examination!
