Create a raw fuzzy system containing only two rules. fis=CreateSystemWith2Rules(InTrain,OutTrain); ------------------------------------------------------------------------- Parameters ------------------------------------------------------------------------- InTrain - matrix containing the input values of the training data points. Each row corresponds to a data point, each column corresponds to an input dimension. OutTrain - column vector containing the output values of the training data points. Each row corresponds to a data point. CreateOptions - column vector containing parameters of the system generation (1) Coefficient for the core (2) Coefficient for the support (3) Apply or not meta rule 1. (4) Coefficient for dmin in meta rule 1. (5) Apply or not meta rule 2. (6) Coefficient for dpmin in meta rule 2. (7) Enable a part of the shape of the linguistic terms to fall outside the range of the linguistic variable. (8) Number of allowed decimals for set parameters. CreateOptions - vector describing the preferences 1 - DecimalsNo - Number of allowed decimals. Defines the precision. 2 - CoreRelWidth - Width of the core of the new sets relative to the range of the linguistic variable. 3 - SupportRelWidth - Width of the support of the new sets relative to the range of the linguistic variable. ------------------------------------------------------------------------- Returned values ------------------------------------------------------------------------- fis - complex structure describing the raw Fuzzy Inference System. ------------------------------------------------------------------------- Remarks ------------------------------------------------------------------------- The two rules aim the description of the minimum and maximum output. First one seeks the two extreme output values and a representative data point for each of them. If several data points correspond to an extreme value, one selects the one that is closer to an endpoint of the input domain. The reference points of the antecedent sets of the first rule will be identical with the corresponding input values of the minimum point. The reference point of the consequent set will be identical with the output value of the minimum point. The shape of the linguistic terms is determined by the default set shape, which is a characteristic feature of the dimension. The antecedent and consequent linguistic terms of the second rule are determined in a similar way taking into consideration the maximum point. At this point the system contains two linguistic terms in each dimension. This version of the function supports only trapezoid shaped fuzzy sets. In order to fulfill the conditions related to trapezoid sets, the widths of the cores and supports are shrunk the same amount if it is necessary. However, their reference points are kept always unmodified. In case of a coincidence in any of the dimensions the sets are unified and only the first specified set is kept. ------------------------------------------------------------------------- Zsolt Csaba Johanyák, johanyak.csaba@gamf.kefo.hu, v. 1.2, 27 Aug 2007.

- CorrectSetShapes Correct set shapes after the introduction of the last set in order to fulfil the conditions of tuning method RBE-DSS.
- MergeSimilarSets Merge fuzzy sets whose params does not differ more than
- RoundFisData
- SortMfs Sort the membership functions in each dimension in ascending order.

Generated on Thu 26-Mar-2009 13:11:50 by