Page 93 - Contributed Paper Session (CPS) - Volume 6

CPS1832 Nur Fazliana Rahim et al.
and allow fuzzy sets to be related with different linguistic variables associated.
The relative weight, for linguistic term , with regard to classification is:
i

where (, )[0,1]and = 1, 2, . . . , . Thus, the compound weight () and

() of the weighted conjunction of linguistic terms related with it can be
obtained as follows;

where A is the conditional attribute and the compound weight is (), ∇ is the
t-norm, , = 1, 2, . . ., , are the linguistic terms of variable which are

conjunctively combined and is the largest amongst the associated
weight, (, ). Similarly, the compound weight () of the weighted

disjunction of linguistic terms associated with variable , where Δ is the t-
conorm and , = 1, 2, . . ., , are the linguistic terms of variable B, which

are disjunctively combined.
Fifth, the WSBA then used the weighted conjunction () and
weighted disjunction () to generate fuzzy rules.
C. Part 3: Finalize the Classification Output
The classification of FER rank can be performed when the rule set and
the study over the three conditional attributes are acquired. Then, using rule
set generated and the transform fuzzy values, the rules were calculated. the
Min-Max Operator is used in this part.
D. Part 4: Rule set Testing for the Classification Tasks
The training dataset, FER-1 using rule set for classification of FER rank
were then tested using the FER-2 dataset. The identified trend for each of the
FER data were based on the classification of the rules. The FER distribution
was started to forecast when each of the FER data had been classified by the
rule and trend of forecasting.

3. Result
The analysis performed in this research were discusses in this section. Table
I below depicted the forecast value obtained by using prior method and the
current approach. The FER-1 dataset are tested using the rule set trained for
classification of Foreign Exchange Rate where the FER-2 dataset is used for
testing. The difference between forecasting methods was illustrates in Fig. 1.

82 | I S I W S C 2 0 1 9