Predicting student grades based on factors present in their everyday lives
The goal of this project is to predict student pass rates and identify the factors that are most intimately correlated to achieving high grades. The dataset used for this paper examines the performance of students and is made up of data from two schools, GP and MS. For each student in the dataset, there is a detailed observation across many variables such as parental status, alcohol intake, future ambitions, parents' job status, health status, and so on.
To predict the student pass rates, three models have been developed, Random Forest (RF), Support Vector Machine (SVM), and Gradient Boosting. The data science lifecycle will be followed for model prediction, starting with data collection and preparation, then exploration and analysis, and lastly evaluating the results of all models developed.
Accuracy, F1 score (involving Precision and Recall), Area Under the Curve (AUC) derived from the receiver operating characteristic curve (ROC), and Matthews Correlation Coefficient (MCC) are the quantitative techniques used to provide a precise evaluation.