Multimodal Learning Analytics (MMLA) researchers are progressively employing machine learning (ML) techniques to develop predictive models to improve learning and teaching practices. These predictive models are often evaluated for their generalizability using methods from the ML domain, which do not take into account MMLA’s educational nature. Furthermore, there is a lack of systematization in model evaluation in MMLA, which is also reflected in the heterogeneous reporting of the evaluation results. To overcome these issues, this paper proposes an evaluation framework to assess and report the generalizability of ML models in MMLA (EFAR-MMLA). To illustrate the usefulness of EFAR-MMLA, we present a case study with two datasets, each with audio and log data collected from a classroom during a collaborative learning session. In this case study, regression models are developed for collaboration quality and its sub-dimensions, and their generalizability is evaluated and reported. The framework helped us to systematically detect and report that the models achieved better performance when evaluated using hold-out or cross-validation but quickly degraded when evaluated across different student groups and learning contexts. The framework helps to open up a “wicked problem” in MMLA research that remains fuzzy (i.e., the generalizability of ML models), which is critical to both accumulating knowledge in the research community and demonstrating the practical relevance of these techniques.

Multimodal learning analytics (MMLA) research has shown the feasibility of building automated models of collaboration quality using artificial intelligence (AI) techniques (e.g., supervised machine learning (ML)), thus enablingthe development of monitoring and guiding tools for computer-supported collaborative learning (CSCL). However, the practical applicability and performance of these automated models in authentic settings remains largely an under-researched area. In such settings, the quality of data features or attributes is often affected by noise, which is referred to as attribute noise. This paper undertakes a systematic exploration of the impact of attribute noise on the performance of different collaboration-quality estimation models. Moreover, we also perform a comparative analysis of different ML algorithms in terms of their capability of dealing with attribute noise. We employ four ML algorithms that have often been used for collaboration-quality estimation tasks due to their high performance: random forest, naive Bayes, decision tree, and AdaBoost. Our results show that random forest and decision tree outperformed other algorithms for collaboration-quality estimation tasks in the presence of attribute noise. The study contributes to the MMLA (and learning analytics (LA) in general) and CSCL fields by illustrating how attribute noise impacts collaboration-quality model performance and which ML algorithms seem to be more robust to noise and thus more likely to perform well in authentic settings. Our research outcomes offer guidance to fellow researchers and developers of (MM)LA systems employing AI techniques with multimodal data to model collaboration-related constructs in authentic classroom settings.