Project – Design and Implementation of an Artificial Intelligence-Based Student Performance Prediction System: A Case Study of Undergraduate Students at the University of Lagos.

CHAPTER ONE

INTRODUCTION

1.1 Background to the Study

The rapid development of Artificial Intelligence (AI) has created new possibilities for improving decision-making and service delivery across different sectors, including education. Artificial Intelligence refers broadly to computational systems capable of performing tasks that traditionally require aspects of human intelligence, such as learning from data, identifying patterns, making predictions and supporting decisions. In higher education, the increasing availability of digital records has created opportunities for institutions to analyse students’ academic activities and identify patterns associated with success or poor performance. Educational data mining and machine learning have consequently become important approaches for transforming large volumes of educational data into useful information for teaching, learning and institutional decision-making. Recent reviews indicate that supervised machine-learning techniques such as decision trees, random forests, support vector machines and logistic regression are increasingly applied to predict student performance and identify learning patterns (Ersozlu, Taheri, & Koch, 2024; Yağcı, 2022). The growing application of these technologies suggests that higher education institutions can move beyond merely recording students’ past performance towards developing systems capable of providing early indications of likely academic outcomes.

Student academic performance is one of the major indicators used to assess the effectiveness of teaching and learning in higher education. It is commonly reflected through students’ grades, grade point averages, course results, progression and completion of academic requirements. However, academic performance is influenced by a combination of academic, behavioural, demographic, social and institutional factors, making it difficult to predict accurately through conventional observation alone. Factors such as previous academic achievement, attendance, assessment results, engagement with learning activities and patterns of participation may contain useful information about a student’s likely future performance. Machine-learning techniques are particularly relevant because they can process multiple variables simultaneously and identify relationships that may not be easily detected through traditional approaches. Yağcı (2022), for example, demonstrated that machine-learning algorithms could be used to predict undergraduate students’ final examination grades from selected academic and institutional variables, while recent systematic evidence shows that machine learning is increasingly being applied to predict performance, engagement and other student outcomes in higher education (Yağcı, 2022; Li et al., 2025).

The development of student performance prediction systems is also connected to the broader movement towards data-driven and personalised education. Rather than waiting until students have failed courses or accumulated poor academic records, predictive systems can potentially provide early information about students who may require additional academic support. Such systems may assist lecturers, academic advisers and administrators in identifying patterns associated with academic difficulty and planning appropriate interventions. A systematic review of deep-learning applications in virtual learning environments found that predictive technologies can help identify students at risk of poor performance and support timely intervention to improve learning outcomes and reduce the possibility of dropout (Alamri et al., 2024). Similarly, the wider literature on machine learning in higher education indicates that predictive analytics is increasingly being connected with student engagement, self-efficacy and academic performance rather than being limited to retrospective analysis of examination results (Li et al., 2025). Consequently, the design of an artificial intelligence-based student performance prediction system can provide a technological mechanism through which educational data may be transformed into actionable academic information.

The usefulness of Artificial Intelligence in education, however, depends on the quality, relevance and responsible use of the data on which predictive systems are developed. A predictive model is only as useful as the information supplied to it, and incomplete, inaccurate or biased educational data may produce unreliable predictions. There are also concerns relating to privacy, transparency, fairness, explainability and the appropriate interpretation of automated predictions. UNESCO (2023) emphasises that the use of AI in education should be human-centred and should protect learners’ rights, agency, inclusion, equity and data privacy. Similarly, recent work on predictive student-performance models has highlighted the growing importance of explainable Artificial Intelligence because educators need to understand why a system has classified a student as likely to perform well or poorly rather than relying blindly on an unexplained algorithmic output (Alamri et al., 2024; UNESCO, 2023). Therefore, the design of a student performance prediction system should not focus solely on prediction accuracy but should also consider data quality, ethical use, interpretability and the practical needs of educational stakeholders.

In the Nigerian higher education environment, the need for data-driven academic support is particularly relevant because universities operate within complex educational environments involving large student populations, diverse programmes, varying academic abilities and substantial demands on lecturers and administrators. Nigerian undergraduates increasingly interact with digital technologies for learning, communication, assessment and information access, thereby generating forms of educational data that can potentially support academic analytics. Recent Nigerian research has shown that Artificial Intelligence is becoming increasingly relevant to undergraduate students and their academic learning behaviour, although important questions remain concerning how AI technologies can be meaningfully integrated into educational processes (Nweke, 2026). The Nigerian context therefore presents an opportunity for the development of locally appropriate AI systems that are designed around the realities of Nigerian universities rather than relying entirely on predictive models developed from foreign educational datasets. Such localisation is important because differences in academic structures, assessment systems, student demographics, technological access and institutional practices may influence the variables that are useful for predicting academic outcomes.

The University of Lagos provides a relevant environment for examining the design and implementation of an AI-based student performance prediction system because it is a major Nigerian university with undergraduate students drawn from different academic disciplines and backgrounds. Undergraduate academic records can contain information such as previous grades, course results, assessment outcomes and other measurable indicators that may be useful for predictive modelling. A properly designed system could potentially analyse such information and generate predictions that support academic monitoring and early intervention. The value of such a system would extend beyond simply producing a predicted grade because it could assist academic advisers and relevant university personnel in recognising students who may need additional support. Research on educational machine learning has shown that predictive models can process academic data and generate useful classifications or predictions that support institutional decision-making (Yağcı, 2022; Ersozlu et al., 2024). However, the effectiveness of such an approach depends on developing and evaluating a model using data and conditions that reflect the institution where the system is intended to operate.

Recent developments in machine learning further demonstrate the feasibility of developing intelligent systems capable of predicting student performance, but they also reveal the need for continuous empirical evaluation. A 2025 systematic review of machine-learning applications in higher education found that prediction of student performance, engagement and self-efficacy has become a significant area of research, with algorithms including decision trees, random forests, support vector machines and neural networks being frequently applied. The review also noted the increasing use of evaluation measures such as accuracy, precision, recall and F1-score, demonstrating that predictive systems must be empirically tested rather than judged only by their technological sophistication (Li et al., 2025). Similarly, Ersozlu et al. (2024) reported that supervised machine-learning methods dominate educational-data applications because of their usefulness in identifying patterns and predicting student outcomes. These developments provide a strong basis for designing and implementing an Artificial Intelligence-based student performance prediction system for undergraduate students at the University of Lagos, while also creating an opportunity to examine the system’s predictive performance within a Nigerian university context.

1.2 Statement of the Problem

Academic performance remains an important concern in higher education because poor academic outcomes may affect students’ progression, graduation, employability and overall educational experience. Universities commonly rely on examination results, continuous assessment scores, cumulative grade point averages and academic advising to monitor student progress. Although these approaches provide valuable information, they are often largely retrospective because meaningful intervention may occur only after a student has already demonstrated poor performance. The growing availability of student academic data creates an opportunity to identify performance patterns before serious academic difficulties become evident. However, without an appropriate analytical system, large quantities of educational data may remain underutilised for predictive and preventive academic decision-making (Yağcı, 2022; Ersozlu et al., 2024).

A second problem concerns the difficulty of accurately identifying students who may experience academic challenges when several factors interact simultaneously. Students’ academic outcomes may be associated with previous academic achievement, assessment results, engagement and other measurable characteristics, and the relationships among these factors may not always be obvious through manual observation. Traditional approaches may therefore provide limited capacity for analysing large and multidimensional datasets to generate timely predictions. Recent research indicates that machine-learning algorithms can identify patterns within educational datasets and use these patterns to predict future academic outcomes, but the effectiveness of individual algorithms varies according to the nature and quality of the data used (Yağcı, 2022; Li et al., 2025). This creates a need for a properly designed system that can process relevant student information and produce reliable predictions that can support academic decision-making.

Another major problem is the limited availability of institution-specific AI-based student performance prediction systems developed around the Nigerian university environment. Much of the existing literature on educational prediction has emerged from datasets and educational systems outside Nigeria, including studies based on universities in Turkey and other international contexts. Although such studies demonstrate the technical feasibility of student performance prediction, their findings cannot automatically be assumed to apply to Nigerian universities because institutional structures, assessment practices, student characteristics and available educational data may differ. Recent Nigerian scholarship confirms that AI is increasingly entering the academic experiences of undergraduate students, but there remains a need to move beyond general discussions of AI use towards practical systems that address specific institutional educational challenges (Nweke, 2026). Therefore, there is a need for empirical work that develops and tests an AI-based prediction system using an appropriate undergraduate student context in Nigeria.

Furthermore, even where Artificial Intelligence is used to support educational prediction, concerns remain regarding prediction accuracy, fairness, privacy, transparency and the practical interpretation of automated results. An inaccurate prediction could potentially lead to inappropriate academic decisions, while an unexplained prediction may be difficult for lecturers or academic advisers to trust and use effectively. UNESCO (2023) stresses that AI applications in education should be implemented in ways that protect privacy, human agency, equity and responsible decision-making. Recent research on student-performance prediction has similarly emphasised the importance of evaluating predictive models using appropriate performance metrics and improving the interpretability of AI outputs (Alamri et al., 2024; Li et al., 2025). The problem, therefore, is not merely the absence of an automated prediction tool but the need for a properly designed, implemented and evaluated system that can use relevant undergraduate academic data to predict student performance with acceptable reliability. It is against this background that this study focuses on the design and implementation of an Artificial Intelligence-based Student Performance Prediction System for undergraduate students at the University of Lagos.

1.3 Purpose of the Study

The main purpose of this study is to design and implement an Artificial Intelligence-based Student Performance Prediction System for undergraduate students at the University of Lagos.

Specifically, the study seeks to:

1. identify the academic and related student variables that can be used in predicting the academic performance of undergraduate students at the University of Lagos;
2. design an Artificial Intelligence-based model for predicting the academic performance of undergraduate students at the University of Lagos;
3. implement an Artificial Intelligence-based Student Performance Prediction System using appropriate machine-learning techniques and relevant student academic data; and
4. evaluate the predictive performance and effectiveness of the developed system using appropriate model-performance measures.

1.4 Research Questions

The following research questions guide the study:

1. What academic and related student variables can be used in predicting the academic performance of undergraduate students at the University of Lagos?
2. How can an Artificial Intelligence-based model be designed to predict the academic performance of undergraduate students at the University of Lagos?
3. How can an Artificial Intelligence-based Student Performance Prediction System be implemented using appropriate machine-learning techniques and relevant student academic data?
4. How effective and accurate is the developed Artificial Intelligence-based Student Performance Prediction System in predicting undergraduate students’ academic performance?

1.5 Research Hypothesis

The following null hypothesis will be tested at 0.05 level of significance:

H₀: There is no significant difference between the academic performance predicted by the developed Artificial Intelligence-based Student Performance Prediction System and the actual academic performance of undergraduate students at the University of Lagos.

1.6 Significance of the Study

The study will be significant to undergraduate students because an effective performance prediction system may provide early indications of their likely academic outcomes. Such information can encourage students to recognise potential academic difficulties and take appropriate corrective measures, including improving their study habits, seeking academic assistance and increasing engagement with their courses.

The study will be significant to lecturers and academic advisers because the developed system may provide additional evidence for monitoring student academic progress. Instead of relying exclusively on retrospective examination results, academic personnel may use predictive information to identify students who require additional academic guidance, mentoring or learning support.

The study will also be useful to university administrators and educational managers. A reliable predictive system can provide an additional data-driven mechanism for understanding patterns of student performance and supporting academic planning. It may assist institutions in developing early-intervention strategies, improving student-support programmes and making more informed decisions concerning academic monitoring.

The study will be beneficial to software developers and information technology professionals because it demonstrates how Artificial Intelligence and machine-learning techniques can be applied to a practical educational problem. The system architecture, data-processing procedures, prediction model and evaluation approach may provide a basis for future educational software development and institutional analytics projects.

The study will also contribute to researchers and scholars by adding to the growing body of literature on Artificial Intelligence, educational data mining, machine learning and predictive analytics in higher education. In particular, the study will provide an institutionally grounded perspective from a Nigerian university, thereby contributing to the localisation of research that has often relied on datasets from other educational environments.

Finally, the study may be useful to policy makers and higher education stakeholders who are interested in responsible AI adoption in education. The findings may provide evidence on the opportunities and limitations associated with using predictive AI for student academic support. This is particularly important because responsible educational AI requires attention to privacy, fairness, transparency and human oversight rather than focusing exclusively on technological capability (UNESCO, 2023).

1.7 Scope of the Study

The study focuses on the design and implementation of an Artificial Intelligence-based Student Performance Prediction System for undergraduate students at the University of Lagos. The study is delimited to the use of relevant student academic and related variables that can be obtained and ethically utilised for predictive modelling.

The study covers the identification of appropriate prediction variables, preparation and processing of relevant student data, selection and development of an appropriate machine-learning model, implementation of the prediction system, and evaluation of its predictive performance. The study will consider appropriate machine-learning techniques for classifying or predicting students’ academic performance based on available data.

Geographically, the study is restricted to the University of Lagos, Lagos State, Nigeria. The study does not attempt to develop a universal prediction model for all Nigerian universities. Rather, it concentrates on developing and evaluating a system within the selected institutional context.

1.8 Operational Definition of Terms

Artificial Intelligence (AI): The capability of computer systems to perform tasks that ordinarily require aspects of human intelligence, including learning from data, pattern recognition, prediction and decision-making.

Academic Performance: The measurable academic outcome achieved by a student, represented in this study through relevant academic indicators such as grades, scores, grade point averages or performance categories.

Artificial Intelligence-Based Student Performance Prediction System: A computer-based system that uses Artificial Intelligence and machine-learning techniques to analyse relevant student data and predict likely academic performance.

Machine Learning: A branch of Artificial Intelligence that enables computer systems to learn patterns from data and use the learned patterns to make predictions or classifications without being explicitly programmed for every individual prediction.

Student Performance Prediction: The process of using historical or current student-related data to estimate or classify a student’s likely future academic outcome.

Educational Data Mining: The application of data-mining and computational techniques to educational data to discover patterns, relationships and information that can support learning and educational decision-making.

Predictive Model: A mathematical or computational model developed from historical data for the purpose of estimating future or unknown outcomes.

Undergraduate Student: A student enrolled in a first-degree academic programme at the University of Lagos.

Prediction Accuracy: The degree to which the output generated by the Artificial Intelligence prediction model corresponds with the actual academic outcome of the student.

Early Intervention: Academic support provided to students identified as potentially experiencing academic difficulty before poor performance becomes more severe.

System Implementation: The process of converting the proposed Artificial Intelligence prediction model and system design into a functional computer-based application capable of receiving data, processing it and generating predictions.

1.9 Organisation of the Study

The study is organised into five chapters.

Chapter One presents the introduction to the study. It covers the background to the study, statement of the problem, purpose of the study, research questions, research hypothesis, significance of the study, scope of the study, operational definition of terms and organisation of the study.

Chapter Two will present the review of related literature. It will discuss the conceptual framework, relevant concepts relating to Artificial Intelligence, machine learning, educational data mining and student performance prediction, empirical studies on student performance prediction, relevant theoretical perspectives and the identified gap in existing literature.

Chapter Three will present the research methodology. It will discuss the research design, study population, sample and sampling technique, sources of data, data collection instruments or data acquisition procedures, system-development approach, data-preparation procedures, machine-learning techniques, system architecture, model training and testing procedures, and methods of data analysis and system evaluation.

Chapter Four will present the results of the study and the implementation of the developed Artificial Intelligence-based Student Performance Prediction System. It will include the presentation and analysis of relevant data, system design, implementation details, model evaluation and interpretation of the findings.

Chapter Five will provide the summary, conclusion and recommendations. It will also highlight the contribution of the developed system, limitations of the study and suggestions for further research.

Project – Design and Implementation of an Artificial Intelligence-Based Student Performance Prediction System: A Case Study of Undergraduate Students at the University of Lagos.