An Application of the Support Vector Machine for Attribute-By-Attribute Classification in Cognitive Diagnosis

Area of Science:

• Educational Measurement
• Psychometrics
• Machine Learning

Background:

• Cognitive diagnostic modeling (CDM) is theoretically advanced but underutilized in practice.
• Challenges include large sample size requirements, computational complexity, and model fit issues.

Purpose of the Study:

• To propose Support Vector Machines (SVM) as a method for cognitive diagnostic modeling.
• To evaluate SVM's performance in attribute classification for educational assessments.

Main Methods:

• Utilized Support Vector Machines (SVM), a supervised learning technique, for attribute classification.
• Transformed CDM fitting and calibration into a quadratic optimization problem.
• Simulated various conditions including sample size, error rates, and CDM parameters.

Main Results:

• SVM achieved classification accuracy comparable to existing methods.
• Effective classification was observed at both attribute and pattern levels.
• The method demonstrated high accuracy even with significantly smaller sample sizes.

Conclusions:

• SVM offers a computationally efficient and accurate approach to cognitive diagnostic modeling.
• This method enhances the practical application of CDM, especially in small-scale testing.
• SVM integration promises to increase the real-world utility of cognitive diagnostic models.