Algorithm 1: Fuzzy Weighted Rule | |
1. | Take the numerical data in tabular form as the input, where the size of each record is N and number of output types are n |
2. | Take M number of MFs for each of N input variable |
3. | Convert numerical input data of the table into Fuzzy linguistic values using MFs |
4. | Take union of Fuzzy linguistic value of 1st field, 2nd field, 3rd field, …, Nth field for the case of first output |
5. | If the sets obtained from the unions of step 4 are: {S1}, {S2}, {S3}, …, {SN} then the N-tuple (({S1}, {S2}, {S3}, …, {SN}), First output) known as RFirst_output |
6. | Repeat step 4 and 5 to get RSecond_output, RThird_output, …, RNth_output |
7. | Take union of 1st element of RFirst_output, RSecond_output, RThird_output, …, RNth_output to get the rule R1 |
8. | Repeat step 7 for 2nd, 3rd, …, Nth elements of RFirst_output, RSecond_output, RThird_output, …, RNth_output to get the rule R2, R3, R4, …, RN |
9. | Take the sum of non-overlapping range and full range of first input variable against all the n output |
10. | Take the ratio v1 two terms of step 9 |
11. | Repeat step 9 and 10 for the rest of input variables |
12. | Take and weights, , where (1) |
13. | For each input record of N-tuple determine weighted co-variance of each rule like, ; (2) where Xj is jth the input Fuzzy variable, i for ith rule, if Xj belongs to jth set of ith rule Ri, otherwise |
14. | The highest value of R corresponding to kth rule indicates the input tuple is under the output of kth category |