When the objective functions are MSDE and MSLE, the MODE-CMCS generates significantly better solutions than NSGA-II.
And two pairs of objective functions, which are MSLE versus MSDE and MSDE versus M4E, are considered.
Caption: Figure 5: Pareto plots of the prediction results when the objective functions are MSLE and MSDE.
Caption: Figure 7: The prediction uncertainty ranges associated with the Pareto optimal solutions generated by MODE-CMCS algorithm when the objective functions are MSLE and MSDE.
Caption: Figure 8: The prediction uncertainty ranges associated with the Pareto optimal solutions generated by MODE-CMCS algorithm when the objective functions are MSLE and MSDE.
Index Problems NSGA-II MODE-CMCS Average Std Average Std GD ZDT1 0.033482 0.004750 0.000319 0.000055 ZDT2 0.072391 0.031689 0.000502 0.000081 ZDT3 0.114500 0.007940 0.000674 0.000073 ZDT4 0.513053 0.118460 0.000120 0.000010 ZDT6 0.296564 0.013135 0.000107 0.000011 SP ZDT1 0.390307 0.001876 0.09014 0.010209 ZDT2 0.430776 0.004721 0.095107 0.007355 ZDT3 0.738540 0.019706 0.126092 0.009318 ZDT4 0.064648 0.100821 0.013693 ZDT6 0.009923 0.077105 0.008142 Table 3 Convergence Performance Of Two algorithms Algorithm GD (MSLE versus MSDE) GD (MSDE versus M4E) NSGA-II 5.21E--2 2.96E--2 MODE-CMCS 2.78E--2 2.55E--2 Table 4: Comparisons of SPC between NSGA-II and MODE -CMCS.
We specified MSLE at 7%, which means the LES must contain event type that have at least been accessed by7%of the users.
While well-known association rule mining algorithms such as Apriori and FP-Growth use minimum support threshold in the generation of rules from a dataset, temporal relationship use the minimum support threshold for large events (MSLE) to generate the large event set, minimum support threshold for uniform event set (MSUE) to generate the uniform event set, and finally minimum support threshold for relation rule (MSRR) to generate the frequent relational rules.
Effect of MSLE. MSLE is used to find large event sets, which represent frequent events in the dataset.
From this figure, we can see that the number of large events generated was highest when smaller MSLE been defined and getting reduced when the MSLE is higher.
We also used the same value of support threshold as MSLE, which is 7% for both algorithms.
Temporal relation rule approach uses several parameters, which hold threshold values, of MSLE, MSUE, and MSRR along the process of rule mining.