Educational data mining and learning analytics have become important research areas for supporting pedagogical analysis, algorithm development, and privacy-preserving educational research. The advancement of natural language processing (NLP) methods in educational contexts depends on the availability of structured and well-documented textual datasets; however, access to real student data is often restricted due to ethical, legal, and privacy concerns. This article presents a fully synthetic textual dataset of student learning habits and preferences generated using a large language model (LLM). The dataset contains 10,000 CSV-formatted records representing fictional students and includes attributes such as education level, study hours, preferred learning methods, learning challenges, motivation levels, opinions on online learning, and primary devices used for study. Data generation was performed using structured prompting strategies with explicitly defined controlled vocabularies to ensure internal consistency and reproducibility while avoiding the use of any real personal information. The resulting dataset follows intentionally controlled and near-uniform distributions, with variables generated under independent constraints. This design limits its suitability for modelling real-world stochastic behaviour or discovering natural correlations but makes it appropriate for benchmarking educational NLP pipelines, evaluating synthetic data generation techniques, and conducting privacy-preserving survey and machine learning experiments.
Early and accurate detection of infectious diseases in cattle is essential for effective animal health management, sustainable livestock production, and global food security. This study develops and validates a deep learning–based diagnostic framework for the automated visual classification of Foot-and-Mouth Disease (FMD) and Lumpy Skin Disease (LSD) in cattle. A custom Convolutional Neural Network (CNN) was designed and trained on a three-class image dataset comprising FMD, LSD, and healthy cattle. To address challenges of class imbalance and field-image variability, the architecture incorporated batch normalization, dropout regularization, and an extensive data augmentation pipeline. The model achieved an overall test accuracy of 74.77% with a macro-averaged F1-score of 0.75, demonstrating robust performance particularly in the recall of LSD (0.92) and FMD (0.77). While its accuracy is lower than some computationally intensive state-of-the-art models, the proposed CNN offers a practical, lightweight, and deployable alternative suitable for resource-limited settings. The results underscore the potential of accessible computer vision tools to serve as preliminary decision-support systems in veterinary diagnostics, enabling timely disease identification, enhanced surveillance, and early intervention in precision livestock farming systems.