Intelligent student performance analytics system using Java APIs, PHP web dashboard, and MySQL: Predictive modeling with explainable AI

Student performance analytics are becoming a major tool in institutions that aims at detecting at-risk learners early in the institution, providing personalized support and enhancing retention and learning outcomes. The literature review brings together studies in the field of educational data mining, learning analytics dashboards, predictive modeling, and explainable artificial intelligence (XAI), and relates the evidence to a working system blueprint through Java APIs, a PHP web dashboard, and a MySQL database. Previous research demonstrates that academic history, demographic background, and, above all, learning management system (LMS) trace data can be meaningfully predicted, however, their results are highly sensitive to the quality of the data and feature engineering, as well as its careful time-dependent validation and compatibility with intervention processes rather than its accuracy per se (Shahiri et al., 2015; Waheed et al., 2020; Bernacki et al., 2020) [8, 9, 38]. Awareness, reflection, and decision-making can be supported with learning analytics dashboards, but it has been shown that to work effectively, predictive and pedagogical elements must be carefully developed and assessed (Matcha et al., 2020; Susnjak et al., 2022) [27, 3]. XAI procedures such as LIME and SHAP can serve to explain the behavior of models locally and globally to assist educators and students in comprehending the models, tracking down spurious trends, and converting forecasts into real actions to build trust, accountability, and adoption of stakeholders (Ribeiro et al., 2016; Lundberg and Lee, 2017; Guidotti et al., 2018) [41, 43]. The review suggests the following reference architecture: data ingestion and orchestration by means of Java REST APIs, risk scores and calibrated probabilities generated by analytics services, explanations and counterfactual advice generated by XAI services, and role-based views provided by a PHP dashboard to the students, faculty, and administrators. The design recommendations are offered at the end of the paper (not only concerning AUC but also concerning calibration and intervention impact, fairness and privacy protection, and sustainable deployment practices in actual learning environments).