Generative Data Imputation for Sparse Learner Performance Data Using Generative Adversarial Imputation Networks

TL;DR

This work tackles missing learner performance data in Intelligent Tutoring Systems by introducing a CNN-enhanced Generative Adversarial Imputation Nets (GAIN) framework adapted to a 3D learner×question×attempt tensor. By combining a mask-driven generator, a hint-based discriminator, and least-squares optimization, the method imputes missing entries without distorting observed data, and it aligns input and output shapes along the learner dimension. Evaluations on AutoTutor ARC, ASSISTments, and MATHia show that the proposed approach outperforms tensor factorization and several GAN baselines in imputation accuracy, while Bayesian Knowledge Tracing confirms that imputed data yield improved parameter estimation and low KL-divergence from original distributions. These results demonstrate the method’s potential to mitigate data sparsity and enable more accurate, adaptive instruction in ITSs, with broad implications for scalable, data-informed learner support. The study thus advances generative imputation in educational data and provides a robust, scalable tool for enhanced learner modeling under sparse data conditions.

Abstract

Learner performance data collected by Intelligent Tutoring Systems (ITSs), such as responses to questions, is essential for modeling and predicting learners' knowledge states. However, missing responses due to skips or incomplete attempts create data sparsity, challenging accurate assessment and personalized instruction. To address this, we propose a generative imputation approach using Generative Adversarial Imputation Networks (GAIN). Our method features a three-dimensional (3D) framework (learners, questions, and attempts), flexibly accommodating various sparsity levels. Enhanced by convolutional neural networks and optimized with a least squares loss function, the GAIN-based method aligns input and output dimensions to question-attempt matrices along the learners' dimension. Extensive experiments using datasets from AutoTutor Adult Reading Comprehension (ARC), ASSISTments, and MATHia demonstrate that our approach significantly outperforms tensor factorization and alternative GAN methods in imputation accuracy across different attempt scenarios. Bayesian Knowledge Tracing (BKT) further validates the effectiveness of the imputed data by estimating learning parameters: initial knowledge (P(L0)), learning rate (P(T)), guess rate (P(G)), and slip rate (P(S)). Results indicate the imputed data enhances model fit and closely mirrors original distributions, capturing underlying learning behaviors reliably. Kullback-Leibler (KL) divergence assessments confirm minimal divergence, showing the imputed data preserves essential learning characteristics effectively. These findings underscore GAIN's capability as a robust imputation tool in ITSs, alleviating data sparsity and supporting adaptive, individualized instruction, ultimately leading to more precise and responsive learner assessments and improved educational outcomes.

Figures (7)

• Figure 1: An illustration for sparse learning performance data in Intelligent Tutoring Systems.
• Figure 2: The proposed GAIN-based imputation architecture for sparse learning performance.
• Figure 3: RMSE comparisons of data imputation models across different attempts for different lessons dataset.
• Figure 4: RMSE changes with varying sparsity increase rates (corresponding to Max Attempts).
• Figure 5: Comparison of BKT parameters (Original vs. Imputed Learning Performance Data) for ASSISTments Lesson 1 (Max Attempt = 4).