k-Means Performance Improvements with Centroid Calculation Heuristics Both for Serial and Parallel Environments

Abstract

k-means is the most widely used clustering algorithm due to its fairly straightforward implementations in various problems. Meanwhile, when the number of clusters increase, the number of iterations also tend to slightly increase. However there are still opportunities for improvement as some studies in the literature indicate. In this study, improved implementations of k-means algorithm with a centroid calculation heuristics which results in a performance improvement over traditional k-means are proposed. Two different versions of the algorithm for various data sizes are configured, one for small and the other one for big data implementations. Both the serial and MapReduce parallel implementations of the proposed algorithm are tested and analyzed using 2 different data sets with various number of clusters. The results show that big data implementation model outperforms the other compared methods after a certain threshold level and small data implementation performs better with increasing k value.

Menu Optimization for Multi-Profile Customer Systems on Large Scale Data

“…We used IBM Ilog Cplex for the numerical solutions of the optimization problems. Data mining, analysis, and user clustering for the data sets have been performed in our previous publications Karimov et al (2015a); Karimov and Ozbayoglu (2015), hence, for details of data analysis we refer the reader to Karimov et al (2015a); Karimov and Ozbayoglu (2015) and focus on the performance evaluations of the optimization of the menu structure.…”

“…Therefore, it is possible that a certain menu item under another menu item can have a higher click count than its parent menu item. Note that mining, clustering, analysis, and classification of ATM usage data has been performed in our earlier studies (Karimov and Ozbayoglu,2015;Karimov et al,2015a), hence, in this paper we do not go into the details of data processing and assume that the user logs are mined in an efficient way (i.e., they are clustered and click counts of menu items for each cluster is known). -Original graph representation of the ATM menu to be optimized is readily available (i.e., actual menu structure in use by the financial establishment serving the customers).…”

“…However, for other menu optimization domains, much larger sets of menu items should be optimized (e.g., web pages) where the use of SOM model is very much needed. In As mentioned, this paper concentrates mainly on the optimization part of generating optimal profile based user menus and the modules related with data mining are described in Karimov and Ozbayoglu (2015), Karimov et al (2015a). Converting this system to incremental framework is an open direction for research.…”

Big Data Meet Cyber-Physical Systems: A Panoramic Survey

“…The clustering approach was based on a combination of Fireworks and Cuckoo-search algorithms with representative points being selected as the centroids. In [117], the authors also used a centroid calculation heuristics to help enhance the clustering performance as the number of clusters increases.…”