Analytical Framework for Revealing Implicit Buyer Characteristics through Data Partitioning Models

• Prof. Luka Petrović

  Faculty of Artificial Intelligence and Market Systems Balkan Advanced Research University Sarajevo, Bosnia and Herzegovina

  Author

Understanding implicit buyer characteristics has become a central challenge in data-driven marketing analytics, particularly in environments where user behavior is distributed across heterogeneous computational and data environments. Traditional segmentation techniques primarily rely on explicit demographic or transactional attributes, which often fail to capture latent behavioral structures embedded in high-dimensional, fragmented, and distributed datasets. This research proposes an analytical framework that leverages advanced data partitioning models, originally developed for heterogeneous high-performance computing (HPC) systems, and adapts them to buyer behavior inference.

The study synthesizes principles from heterogeneous system optimization, load balancing, and performance-aware partitioning strategies (Lastovetsky and Reddy, 2007; Khaleghzadeh et al., 2018; Clarke et al., 2012) to construct a computational framework capable of revealing hidden behavioral clusters. These models, traditionally used for optimizing matrix computations and data-parallel workloads, are repurposed to segment buyer datasets in a manner that preserves behavioral locality and minimizes informational loss. The proposed framework integrates functional performance modeling with adaptive partitioning logic to dynamically assign behavioral data segments across computational nodes, thereby improving clustering precision.

In addition, the framework draws conceptual parallels between customer segmentation in machine learning and workload partitioning in heterogeneous systems, extending findings from advanced clustering research in behavioral analytics (Jatav et al., 2025) to enhance latent pattern discovery. The integration of hierarchical partitioning mechanisms allows for scalable inference of buyer attributes such as purchasing intent, responsiveness, and preference volatility.

Empirical synthesis of prior HPC partitioning methodologies suggests that data-aware distribution significantly enhances pattern recognition accuracy in distributed analytics systems (Becker, 2010; Hoefler et al., 2014). Building upon these insights, the proposed analytical framework demonstrates that optimizing data locality not only improves computational efficiency but also enhances semantic clarity in buyer segmentation outputs.

The study contributes a unified perspective that bridges high-performance computing and marketing analytics, offering a scalable, model-driven approach for uncovering implicit buyer characteristics in large-scale, heterogeneous datasets. The findings underscore the importance of partition-aware analytics in next-generation intelligent marketing systems.

B. A. Becker, “High-level data partitioning for parallel computing on heterogeneous hierarchical computational platforms by,” Ph.D. dissertation, Dept. Comput. Sci. Inform., College Eng. Math. Phys. Sci., Univ. College Dublin, Dublin, Ireland, 2010.

B. Bastem, D. Unat, W. Zhang, A. Almgren, and J. Shalf, “Overlapping data transfers with computation on GPU with tiles,” in Proc. Int. Conf. Parallel Process., 2017, pp. 171–180.

D. Clarke, A. Lastovetsky, and V. Rychkov, Column-Based Matrix Partitioning for Parallel Matrix Multiplication on Heterogeneous Processors Based on Functional Performance Models (Lecture Notes in Computer Science, Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7155, no. 1. Berlin, Germany : Springer, 2012, pp. 450–459.

A. DeFlumere, A. Lastovetsky, and B. A. Becker, “Partitioning for parallel matrix-matrix multiplication with heterogeneous processors: The optimal solution,” in Proc. IEEE 26th Int. Parallel Distrib. Process. Symp. Workshops PhD Forum, May 2012, pp. 125–139.

Y. Gao and P. Zhang, “A survey of homogeneous and heterogeneous system architectures in high performance computing,” in Proc. IEEE Int. Conf. Smart Cloud (SmartCloud), Nov. 2016, pp. 170–175.

T. Hoefler, E. Jeannot, and G. Mercier, “An overview of topology mapping algorithms and techniques in high-performance computing,” High-Performance Computing on Complex Environments, vol. 9781118712. Hoboken, NJ, USA : Wiley, 2014, pp. 73–94.

H. Khaleghzadeh, R. R. Manumachu, and A. Lastovetsky, “A hierarchical data-partitioning algorithm for performance optimization of data-parallel applications on heterogeneous multi-accelerator NUMA nodes,” IEEE Access, vol. 8, pp. 7861–7876, 2020.

H. Khaleghzadeh, R. R. Manumachu, and A. Lastovetsky, “A novel data-partitioning algorithm for performance optimization of data-parallel applications on heterogeneous HPC platforms,” IEEE Trans. Parallel Distrib. Syst., vol. 29, no. 10, pp. 2176–2190, Oct. 2018.

A. Kalinov and A. Lastovetsky, “Heterogeneous distribution of computations solving linear algebra problems on networks of heterogeneous computers,” J. Parallel Distrib. Comput., vol. 61, no. 4, pp. 520–535, Apr. 2001.

Lastovetsky, L. Szustak, and R. Wyrzykowski, “Model-based optimization of EULAG kernel on Intel Xeon phi through load imbalancing,” IEEE Trans. Parallel Distrib. Syst., vol. 28, no. 3, pp. 787–797, Mar. 2017.

Lastovetsky and R. Reddy, “Data distribution for dense factorization on computers with memory heterogeneity,” Parallel Comput., vol. 33, no. 12, pp. 757–779, Dec. 2007.

Lastovetsky and R. Reddy, “Data partitioning with a functional performance model of heterogeneous processors,” Int. J. High Perform. Comput. Appl., vol. 21, no. 1, pp. 76–90, Feb. 2007.

Lastovetsky and R. Reddy, “Data partitioning with a realistic performance model of networks of heterogeneous computers with task size limits,” in Proc. 3rd Int. Symp. Parallel Distrib. Comput./3rd Int. Workshop Algorithms, Models Tools Parallel Comput. Heterogeneous Netw., 2004, pp. 133–140.

Lastovetsky and R. Reddy Manumachu, “New model-based methods and algorithms for performance and energy optimization of data parallel applications on homogeneous multicore clusters,” IEEE Trans. Parallel Distrib. Syst., vol. 28, no. 4, pp. 1119–1133, Apr. 2017.

Lastovetsky and R. Reddy, Two-Dimensional Matrix Partitioning for Parallel Computing on Heterogeneous Processors Based on Their Functional Performance Models (Lecture Notes in Computer Science, Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6043. Berlin, Germany : Springer, 2010, pp. 112–121.

Meade, D. K. Deeptimahanti, J. Buckley, and J. J. Collins, “An empirical study of data decomposition for software parallelization,” J. Syst. Softw., vol. 125, pp. 401–416, Mar. 2017.

O. Beaumont, V. Boudet, F. Rastello, and Y. Robert, “Matrix-matrix multiplication on heterogeneous platforms,” in Proc. Int. Conf. Parallel Process., 2000, pp. 289–298.

Y. Ogata, T. Endo, N. Maruyama, and S. Matsuoka, “An efficient, model-based CPU-GPU heterogeneous FFT library,” in Proc. IEEE Int. Symp. Parallel Distrib. Process., Apr. 2008, pp. 1–10.

P. Bhat, V. Prasanna, and C. S. Raghavendra, “Block-cyclic redistribution over heterogeneous networks,” Cluster Comput., vol. 3, no. 1, pp. 25–34, 2000.

TOP500 Supercomputer Sites. Accessed: 2019. [Online]. Available: http://top500.org/

Hamidreza Khaleghzadeh_HPOPT GitLab. Accessed: 2020. [Online]. Available: https://csgitlab.ucd.ie/HKhaleghzadeh/hpopt

D. S. Jatav, M. H. Mirza, M. Pal, A. Tripathi and R. Nair, "Uncovering Latent Behavioral Patterns Using Advanced Clustering in Customer Segmentation," 2025 IEEE International Conference on Advanced Computing Technologies (ICACT), Tirupati, India, 2025, pp. 590-595, doi: 10.1109/ICACT67549.2025.11351402.

S. Tabik, G. Ortega, E. M. Garzón, and D. Suárez, “A data partitioning model for highly heterogeneous systems,” in Proc. Euro-Par Parallel Process. Workshops Euro-Par, vol. 10104, F. Desprez, Eds. Cham, Switzerland : Springer, 2017, pp. 468–479.

D. Unat, “Trends in data locality abstractions for HPC systems,” IEEE Trans. Parallel Distrib. Syst., vol. 28, no. 10, pp. 3007–3020, Oct. 2017.

D. Unat, T. Nguyen, W. Zhang, and M. N. Farooqi, “TiDA: High-level programming abstractions,” in Proc. Int. Conf. High Perform. Comput., vol. 1. Cham, Switzerland : Springer, 2016, pp. 116–135.

D. Unat, T. Schulthess, T. Hoefle, A. Dubey, and J. Shalf, “Programming abstractions for data locality,” in Proc. PADAL Workshop Swiss Nat. Supercomput. Center (CSCS), Lugano, Switzerland, Apr. 2014.

A. Snavely, X. Gao, C. Lee, L. Carrington, N. Wolter, J. Labarta, J. Gimenez, and P. Jones, “Performance modeling of HPC applications,” in Advances in Parallel Computing, vol. 13. Amsterdam, The Netherlands : North Holland, 2004, pp. 777–784.

A. Yang, F. Wang, Y. Du, J. Chen, J. Liu, H. Yi, and K. Lu, “Adaptive optimization for petascale heterogeneous CPU/GPU computing,” in Proc. IEEE Int. Conf. Cluster Comput., Sep. 2010, pp. 19–28.

• PDF

Copyright (c) 2026 Prof. Luka Petrović (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.