This repository presents a research project that applies Cluster Analysis (K-Means) to large-scale educational assessment data.
The study proposes an alternative methodology for analyzing educational performance using clustering techniques, providing a more representative view of student development than the traditional evaluation strategy adopted by the assessment system.
The work is based on a peer-reviewed scientific publication published in EDUCA – Multidisciplinary Journal of Education (2020). :contentReference[oaicite:0]{index=0}
- Overview
- Problem
- Objectives
- Dataset
- Methodology
- Results
- Repository Contents
- Data Availability
- Scientific Publication
- Citation
- Author
- License
Large-scale educational assessment systems are widely used to evaluate educational quality and support public policies.
However, traditional evaluation methods may not fully capture the real patterns of student development.
This study investigates whether Cluster Analysis can provide a more reliable representation of educational performance than the methodology traditionally adopted by the SPAECE assessment system. :contentReference[oaicite:1]{index=1}
- Apply Cluster Analysis to educational assessment data.
- Compare the proposed methodology with the official SPAECE evaluation approach.
- Identify meaningful groups of municipalities with similar educational performance.
- Support evidence-based educational policy decisions.
The research uses official data from SPAECE 2014 (Permanent System of Educational Assessment of Ceará).
The analysis includes:
- Ceará State educational assessment data
- Mathematics and Portuguese performance
- 2nd, 5th and 9th grade students
- Large-scale educational assessment
:contentReference[oaicite:2]{index=2}
The proposed methodology combines statistical analysis and unsupervised learning techniques.
Main steps:
- Data collection
- Data preprocessing
- Cluster Analysis
- K-Means clustering
- Comparative analysis
- Educational interpretation of the clusters
The analyses were originally conducted using IBM SPSS Statistics.
The comparative analysis indicates that the proposed clustering methodology provides a more accurate representation of educational development than the traditional SPAECE evaluation approach.
The study concludes that these results should be considered in the development of educational public policies. :contentReference[oaicite:3]{index=3}
educational-clustering-analysis/
├── images/
│ ├── banner-project.png
│ └── README.md
│
├── paper/
│ ├── article.pdf
│ └── README.md
│
├── README.md
├── LICENSE
└── .gitignore
The original research data are derived from institutional educational assessment databases.
The original analytical project files are not publicly available.
This repository provides:
- Project documentation
- Scientific publication
- Research overview
- Methodological description
This repository is based on the peer-reviewed publication:
Clustering Method Applied to Educational Assessment: A Case Study for Large-Scale Assessments
EDUCA – Multidisciplinary Journal of Education
Volume 7 (2020)
DOI:
https://doi.org/10.26568/2359-2087.2020.4413
📄 Published Article
@article{Pereira2020,
title={Clustering Method Applied to Educational Assessment: A Case Study for Large-Scale Assessments},
author={Pereira, Valberto Rômulo Feitosa and Paula, Anderson Damasceno de and Araújo, Cristian Oliveira},
journal={EDUCA – Revista Multidisciplinar em Educação},
volume={7},
year={2020},
doi={10.26568/2359-2087.2020.4413}
}Data Scientist | Applied Statistics | Machine Learning | Educational Analytics
Professor and Researcher at the Federal Institute of Education, Science and Technology of Ceará (IFCE).
- GitHub: https://github.com/ValbertoFeitosa
- LinkedIn: https://www.linkedin.com/in/valberto-feitosa-7239511b1/
This project is distributed under the MIT License.