As mentioned in the previous sections, k-means is computationally light and converges after a limited iterations. However, the studies conducted by the researchers confirm that the algorithm is highly dependent on the initialization of centroids and usually gets stuck in local optimums. It was also mentioned that the ABC algorithm performs a global search in the entire solution space. If given enough time, ABC can generate good and global results. Therefore, here, we propose a new combined algorithm to use the merits of the two k-means and ABC algorithms for solving clustering problems. The proposed algorithm does not depend on the initial centroids and can avoid being trapped in a local optimum solution as well.