Scholarlink

<p>This study aims to analyse structural change (SC) patterns across 147 countries/regions from 2000–2021. The research questions are how many patterns of SC can be identified and what the characteristics of these countries’/regions’ patterns are. These research questions were addressed by applying unsupervised machine learning techniques (MLTs) such as principal component analysis (PCA) and cluster analysis (CLA), grounded in a systems approach to the SC, which posits that the SC process emerges from the interaction of system components, undergoing qualitative and quantitative alterations manifested in a subjacent tendency toward an SC or its constraint. The PCA allowed the identification of two dynamic latent dimensions that capture the long-term tendencies of the variables examined, based on which the SC patterns were inferred. These latent dimensions are human development, synergic complementarities, complexity and diversification progress (HDSCD), and structural change and institutional enhancement (SCI). Four SC patterns were identified: Pattern 1 comprises countries/regions that undergo systemic evolution in both latent dimensions; Pattern 2 encompasses countries/regions exhibiting a systemic decline or stagnation in the SCI and a deceleration in the HDSCD; Pattern 3 characterises countries with a systemic trend in the SCI advancement, but a deceleration in the HDSCD; and Pattern 4 comprises countries displaying a decline in the SCI but a systemic evolution in the HDSCD. These patterns were subsequently contrasted with the trends in reducing gender inequalities and planetary pressures through a CLA. The CLA revealed three clusters according to SC patterns and gender and environmental results, providing factual evidence, challenges and development guidelines for each group of countries/regions. These results provide insight into the trajectories of countries/regions with comparable problems, elucidating potential lessons from countries/regions at different development levels to achieve sustainable development goals (SDGs).</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Unveiling structural change patterns: an unsupervised machine learning approach to long-term dynamics

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月16日

查看全文
  • Julián Salinas

摘要

This study aims to analyse structural change (SC) patterns across 147 countries/regions from 2000–2021. The research questions are how many patterns of SC can be identified and what the characteristics of these countries’/regions’ patterns are. These research questions were addressed by applying unsupervised machine learning techniques (MLTs) such as principal component analysis (PCA) and cluster analysis (CLA), grounded in a systems approach to the SC, which posits that the SC process emerges from the interaction of system components, undergoing qualitative and quantitative alterations manifested in a subjacent tendency toward an SC or its constraint. The PCA allowed the identification of two dynamic latent dimensions that capture the long-term tendencies of the variables examined, based on which the SC patterns were inferred. These latent dimensions are human development, synergic complementarities, complexity and diversification progress (HDSCD), and structural change and institutional enhancement (SCI). Four SC patterns were identified: Pattern 1 comprises countries/regions that undergo systemic evolution in both latent dimensions; Pattern 2 encompasses countries/regions exhibiting a systemic decline or stagnation in the SCI and a deceleration in the HDSCD; Pattern 3 characterises countries with a systemic trend in the SCI advancement, but a deceleration in the HDSCD; and Pattern 4 comprises countries displaying a decline in the SCI but a systemic evolution in the HDSCD. These patterns were subsequently contrasted with the trends in reducing gender inequalities and planetary pressures through a CLA. The CLA revealed three clusters according to SC patterns and gender and environmental results, providing factual evidence, challenges and development guidelines for each group of countries/regions. These results provide insight into the trajectories of countries/regions with comparable problems, elucidating potential lessons from countries/regions at different development levels to achieve sustainable development goals (SDGs).