Clustering Models for Sustainable Development Indicators

Clarissa Anaid Parada Cereceres

Diplomová práce

Abstract:

With the 2030 approaching, the Sustainable Development Goals (SDGs) are urgent and crucial in guiding global actions towards a fairer and more sustainable future. Through data mining, global development data can be transformed into meaningful knowledge to guide strategic interventions. Among the data mining techniques, clustering allows the exploration of SDG data without requiring predefined categories, making it perfect for discovering natural groupings among countries or regions. Therefore, the reliability of clustering models is key, as inaccurate groupings may misrepresent progress, misguide resource allocation, and lead to ineffective interventions. This thesis explores clustering techniques: K Means, DBSCAN and Agglomerative Hierarchical Clustering, to detect patterns within selected SDGs. Following the CRISP-DM framework, datasets for three SDGs were prepared, clustered, and evaluated using Silhouette Score, Within-Cluster Sum of Squares, Davis-Boulding and Dunn Index. The results highlighted strengths and limitations of each clustering model among the three datasets. K-Means performed better in two of three datasets, especially when more countries and variables were present, though it was sensitive to outliers and missing values. DBSCAN exhibited a high tendency to assign observations to inappropriate clusters according to its Silhouette Scores, while agglomerative clustering showed moderate performance, outperforming K-Means in one case. The findings emphasize the importance of aligning clustering methodologies with data characteristics and analytical objectives.

Keywords

Clustering models, Development Indicators, Country, K-Means, DBSCAN, Agglomerative Clustering, CRISP-DM, Silhouette Score

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 30. 3. 2025

Obhajoba závěrečné práce

• Obhajoba proběhla 5. 6. 2025
• Vedoucí: doc. Ing. Tomáš Hlavsa, Ph.D.
• Oponent: Michal Kadlec, externi

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Magisterský studijní program:
Informatics