A multidimensional analysis of the 21^st century competencies scale through ai-driven data mining techniques

Area of Science:

• Educational research
• Computer science
• Data science

Background:

• Evaluating student competencies like knowledge, skills, character, and meta-learning is crucial in education.
• Large and complex educational datasets necessitate advanced analytical tools.
• Machine learning (ML) offers powerful methods for analyzing student responses and supporting assessments.

Purpose of the Study:

• To classify student responses from the 21st Century Competencies Scale using various ML algorithms.
• To evaluate the performance of ML models in assessing different competency dimensions.

Main Methods:

• Utilized a dataset of 616 participants' responses across four key sub-dimensions.
• Applied multiple ML algorithms: Support Vector Machine (SVM), Artificial Neural Network (ANN), k-Nearest Neighbors (k-NN), Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), AdaBoost, Gradient Boosting, and XGBoost.
• Optimized model performance using grid search and evaluated using accuracy, precision, recall, and F1-score.

Main Results:

• Logistic Regression (LR) achieved the highest accuracy (78.73%) for the "Character" sub-dimension.
• Support Vector Machine (SVM) showed high accuracy for the "Skills" (78.58%) and overall scale (74.51%) dimensions.
• Gradient Boosting and k-NN models demonstrated competitive performance across multiple dimensions.

Conclusions:

• Machine learning models, especially when optimized, are effective tools for classifying student competencies.
• These findings support the integration of ML in data-driven educational assessment processes.
• The study highlights the potential of ML to provide nuanced insights into student capabilities.