In this paper, we address the important problem of student academic performance prediction, with an emphasis on improving the learning process for lowering failure rates. This research has combined several algorithms such as Logistic Regression Naive bayes, Decision Tree, K-Nearest Neighbor (KNN), Random Forest (RF) and Support Vector Machine (SVM). Additionally, feature engineering is performed to make the data more understandable to machine learning models. The dataset in tabular form includes various information of the students (familial background, ID, gender, courses, marks etc.) and an extra column is added, which stands for the attribute of average success rate or course mark percentage. Based on these lines of evidence, we investigate the importance of predicting the students' current state and future academic statuses. 'If prevention is better than a cure' then we aim to support teachers to intervene and support young people appropriately. The analysis attempts to find such dependencies for final exams, suggesting which courses to enrol in the coming semester according to personal needs. In so doing, we mitigate the universal problem that students fall behind in the absence of personalized guidance and monitoring. This is the big idea behind this research, which ultimately maximizes prediction accuracy at 69.2% and aims to empower learners as well as educators through the use of artificial intelligence to make timely inferences about student academic pathways. By anticipating future outcomes and providing actionable suggestions, our framework strives to preemptively help students to avoid bad things, thus improving a caring and responsive educational environment.