A Hybrid Artificial Bee Colony Algorithm using Multiple Linear Regression on Time Series Datasets

Heuristic algorithms are successfully being applied to solution of time series datasets. The algorithms can be improved further for an elevated level of success. In this study, a hybrid application of the successful artificial bee colony (ABC) algorithm with the statistical multiple linear regression (MLR) method is presented. The proposed algorithm is applied to 3 benchmark time series datasets commonly used in the literature, and favorable results are obtained compared to other similar studies. Pure multiple linear regression, on the other hand, is shown to be unable to reach the same level of success. The results demonstrate that the hybrid application of ABC and multiple linear regression produces satisfactory results in time series datasets.