Hierarchical Clustering and Prediction Optimization Algorithm for Heterogeneous Panel Data

Yu, Xiangjun

doi:10.1142/s0218001426590184

Return

Hierarchical Clustering and Prediction Optimization Algorithm for Heterogeneous Panel Data

2026-04-01

PRE

Yu, Xiangjun

Xin, Ling

Fan, Xiaodi *

DOI：10.1142/s0218001426590184

Original

Original request for help

Save

Abstract

En 中文

In the analysis of heterogeneous panel data, such as data from economics where information is collected across different time periods and individuals, industries, or regions, existing clustering and prediction techniques often fail to capture the underlying complex structures inherent in the data (for example, differences in behavior, trends, or correlations between variables across subgroups, such as varying economic conditions in different sectors or regions). The core research problem addressed by this paper is how to effectively model and predict outcomes from such diverse datasets, where traditional methods often struggle with capturing multi-level relationships and heterogeneity. This paper introduces the HCPHO algorithm, a novel hierarchical clustering approach combined with an adaptive prediction optimization strategy. The algorithm first partitions the data into multiple levels, allowing for the discovery of diverse patterns across different segments. Subsequently, it employs a dynamic optimization process to improve prediction accuracy by adjusting model weights and selecting relevant features based on local and global data characteristics. Experimental results demonstrate that HCPHO outperforms traditional methods in terms of both clustering accuracy and prediction performance, making it a valuable tool for heterogeneous data analysis in various domains.

Keywords:

Hierarchical clustering

prediction optimization

heterogeneous panel data

adaptive learning

data segmentation