How can machine learning personalize educational content generated by LLMs

Modeling Student Retention and Success Using Explainable Machine Learning Techniques

This report presents a reproducible study of student dropout and score prediction using the Open University Learning Analytics Dataset (OULAD) and the code, experiments and outputs contained in the provided Jupyter notebook. The OULAD tables were joined and student-level features were engineered by aggregating Virtual Learning Environment (VLE) interactions (total clicks, active days, distinct resources) and linking demographic and assessment records to each student instance [1]. An 80/20 stratified train–test split produced 32,640 training rows and 8,161 test rows, and preprocessing was implemented via a Column Transformer pipeline: numeric imputation and scaling plus one-hot encoding for categorical fields. To address class imbalance the training set was synthetically balanced with SMOTE producing a balanced class distribution (17,996 examples per class) prior to model training [2], using the imbalanced-learn implementation called from the notebook [4]. A consistent evaluation helper computed accuracy, weighted F1, precision, recall and produced normalized confusion matrices for each fitted model; model explanations for the Elastic Net logistic model were explored with a SHAP demo on a small test subset [5]. The notebook trains the following classifiers (as implemented with scikit-learn): Elastic Net Logistic Regression, Gaussian Naive Bayes, K-Nearest Neighbors (k=7, distance weights), SVM (RBF kernel), MLP Classifier (two hidden layers 128,64 with early stopping) and a soft voting ensemble combining selected members [3]. Reported test set performances (from the notebook runs) are: Gaussian NB — Accuracy 74.94%; KNN — 77.98%; Elastic Net Logistic Regression — 80.88%; SVM (RBF) — 83.81%; MLP Classifier — 84.93%; Soft Voting Ensemble — 84.21%. Normalized confusion matrices and classification reports for each model are displayed in the notebook and used to compare per-class recall and precision. Taken together, the notebook shows that (1) preprocessing with careful aggregation of VLE interactions plus class rebalancing materially improves classifier performance on OULAD-derived targets; (2) a tuned MLP achieved the highest single-model accuracy on the test split (≈84.9%); and (3) model interpretability was briefly demonstrated using SHAP for the logistic model to surface feature contributions on a test subset. All experiments, metrics, plots (including confusion matrices) and numeric values reported here are reproduced directly from the executed notebook cells and their outputs; external references below cite the dataset and the primary tool / method papers used or referenced in the notebook.

Student Academic Performance Prediction Using Machine Learning

The fast growth of educational data systems has led to more student data becoming available at scale to use in learning analytics. It is important to effectively analyze these data to forecast academic performance, as this will facilitate early detection of risks and individualized interventions. The present paper explores some of the most important determinants of academic achievement and assesses several predictive models based on the Student Performance Factors (SPF) dataset (N = 6,607). Linear Regression (LR) and Random Forest (RF) models are built and benchmarked against each other. As can be seen, the RF model, according to the results (R² = 0.70), is significantly outperforming the LR model (R² = 0.62), and the error rate has been minimized by 11.5 percent. Feature importance analysis indicates that attendance is the main determinant (importance weight = 0.381), followed by study hours (0.243) and past scores (0.091). Interestingly, the combination of the existing learning behaviors is six times greater than the historical performance, and the family background factors have insignificant direct effects. The results obtained can be used to justify data-driven educational interventions and imply that the monitoring of attendance should become the central element of any academic early warning system.

AI-Powered Student Dropout Prediction and Personalized Intervention Using TC-Net in Education

Increasing student dropout rates, caused by social, academic, and personal obstacles, raise significant concerns for educational systems. This research introduces an AI-based method for the early detection and support of at-risk students utilizing the Student Final Grade Prediction dataset from two schools in Portugal. Following data preprocessing, which involved cleaning and feature extraction via t-distributed Stochastic Neighbor Embedding (t-SNE), a hybrid model integrating Tabular Data Network and Capsule Networks (TC-Net) was created for performance forecasting. The model obtained a Mean Squared Error (MSE) of 0.43 and an R-squared value of 60.57%, demonstrating robust predictive accuracy. Personalised strategies were then implemented, targeting 90% of at-risk students. Specifically, 80% participated in tailored learning plans, and 70% accessed tutoring support. The findings demonstrate the capability of AI models in minimising dropout risks and enhancing academic performance through prompt, data-informed assistance.

Student Performance Prediction using OLAMLTs: A Machine Learning Approach

LMSs (Learning management systems) are widely used in educational institutions. They are software systems for cloud-based training that are offered locally, remotely, and on demand. As technological costs in higher education decrease, a significant obstacle to online learning is the high cost of creating its content. LMSs are very helpful to education during pandemics period. which significantly impacted worldwide education. In these situations, using LMSs in education offers clever substitutes for traditional classroom instruction and enables teachers to give specialized information, make use of different pedagogical approaches, and better engage their students in their studies. Globally, the current epidemic has caused unanticipated and quick transitions to remote learning and instruction, changes in both content and character. Academic performance largely depends on students' capacity to adapt and respond to disturbances. By focusing on how adaptability contributes to students' educational development and online learning, this study aims to identify factors affecting the adaptability of students in online learning scenarios. For these examinations, this paper suggests using MLTs (Machine Learning Techniques). The OLAMLTs (Online Learner Adaptability Assessment based on MLTs) suggested method evaluates aspects that influence online learners' adaptabilities to education while making recommendations for enhancements, and the recommendations achieve the highest classification accuracy when compared with other methods in evaluations.

An ML Ensemble Model for Evaluating Student Performance and Recommending Academic Specializations in Science or Literature

RoCulturaMCQ: Building a Benchmark While Learning Statistics

The broad adoption of Large Language Models (LLMs) has increased the need for human-curated datasets that serve as evaluation benchmarks. This need is particularly pronounced for non-English languages and for tasks that are inherently subjective and require multiple human perspectives. One such example is the development of benchmarks designed to assess the cultural awareness of LLMs. Statistics and data science courses offer a potential setting for developing such benchmarks while teaching students to apply LLM evaluation techniques using statistical inference. This paper presents a pilot project in which students in a statistics course within a data science engineering program created culturally diverse multiple-choice questions, generated answers using LLMs, and applied statistical methods to assess model accuracy. Student feedback indicated the project was engaging and useful for learning, while also highlighting a notable reliance on LLMs, particularly for interpreting statistical results. The resulting dataset comprises 1355 multiple-choice questions across 18 categories, including language, social media, and politics. After filtering valid items, the dataset was used to evaluate both closed- and open-source LLMs. Results show that the Gemini (closed-source) and Qwen (open-source) model families achieved the best performance, with improvements linked to model size, reasoning capabilities, and access to search tools. The best closed-source model achieved an accuracy of 97.66%, whereas the best open-source model achieved an accuracy of 79.07%. Qualitative analyses of errors in the filtering procedure and model reasoning process point to possible explanations into the challenges LLMs face when handling culturally specific content. Furthermore, results support a cultural injection hypothesis, whereby cultural knowledge is embedded during pretraining and accessed through instruction tuning. Through this work, we aim to demonstrate how statistics and data science courses can provide productive contexts for developing open-source benchmarks for non-English languages while also enriching students’ learning experiences. The dataset is publicly available.