AI-powered Personalized Learning Analytics Platform for Computer Education

Project Abstract

This study presents the design, implementation, and evaluation of an AI-powered personalized learning analytics platform tailored for computer education, aimed at enhancing student outcomes, engagement, and self-regulated learning. The platform integrates multi-modal data from learning management systems, coding practice environments, and assessment results to construct a holistic view of each learner’s strengths, misconceptions, and progression. Leveraging advanced algorithms in adaptive learning, anomaly detection, and knowledge tracing, the system dynamically generates personalized learning paths, provides timely feedback, and recommends targeted resources. A hybrid data pipeline combines real-time activity streams with historical performance to update learner models continuously, while privacy-preserving techniques ensure data security and compliance with ethical standards. The research deploys a mixed-methods approach comprising three phases (1) system development and validation, including a modular architecture with data ingestion, feature extraction, model training, recommendation engine, and visualization dashboards; (2) a quasi-experimental effectiveness study involving multiple undergraduate computer science cohorts to compare learning gains, engagement metrics, and retention between the AI-powered platform and traditional instructional methods; and (3) a qualitative evaluation through interviews and focus groups with students and instructors to assess usability, perceived fairness, and instructional impact. Quantitative outcomes focus on knowledge mastery measured by standardized assessments, skill proficiency evidenced by coding task performance, time-on-task, persistence in problem-solving, and reduction in achievement gaps across diverse learner groups. Effectiveness analyses employ hierarchical linear modeling, propensity score matching, and effect size computation to account for nested data structures and potential confounders. Key contributions include (a) a robust knowledge tracing model capable of handling partial and noisy data from coding exercises, (b) a personalized recommendation system that aligns challenges and resources with both cognitive readiness and affective states, (c) a transparent explainable AI module that surfaces justification for guidance to learners and instructors, (d) a privacy-by-design framework with differential privacy and federated learning options, and (e) a scalable deployment strategy that supports curricular alignment, cross-platform interoperability, and teacher augmentation through analytics dashboards. The platform also supports formative assessment orchestration, enabling timely adjustments to instructional sequences and assessment timing to optimize cognitive load and mastery trajectories. Preliminary findings indicate significant improvements in concept retention, problem-solving efficiency, and course satisfaction for participants exposed to the AI-driven personalized pathway compared with control groups, with notable gains among underrepresented and novice programmers. The study discusses implications for pedagogy, curriculum design, and institutional policy, while addressing limitations such as potential biases in data collection, the need for ongoing model recalibration, and considerations for equitable access across diverse educational settings. Recommendations are provided for practitioners and researchers aiming to scale intelligent learning analytics in computer education without compromising ethical standards or learner autonomy.

Project Overview

What This Project Is About

---

A straightforward exploration of how data about a student’s learning activity can be used to tailor computer education. The project builds a simple system that collects information on how students interact with learning materials, analyzes patterns, and suggests personalized actions to improve learning outcomes.

---

---

The Problem It Addresses

---

Many computer education programs use a one-size-fits-all approach, which may not meet each student’s needs. Without feedback on how students learn, instructors cannot easily identify who needs help or which teaching methods work best for different learners.

---

---

Objectives of the Project

---

1. Identify key learning activities that indicate understanding or confusion.
2. Design a simple analytics workflow to surface personalized recommendations.
3. Prototype a user-friendly interface for students and instructors.

---

---

What You Will Do Step by Step

---

1. Review basic theories of learning and analytics in education.
2. Collect anonymized interaction data from a mock or existing learning platform.
3. Analyze data to find patterns linked to performance improvements.
4. Develop lightweight rules for personalized prompts and resources.
5. Build a basic prototype dashboard to display insights.
6. Test with a small group of users and gather feedback.

---

---

Expected Outcome

A functional prototype that provides personalized study tips and resource suggestions based on student activity, plus a report on what features helped learning most and how it could be improved for broader use.