The aim of this research was to discover how well-informed the people of Bangladesh are when it comes to malicious software and how much of an awareness level they have of the problem. The system is able to examine usage patterns and trends in awareness by applying supervised machine learning technologies. The research used a comprehensive dataset obtained using a Google Forms survey to evaluate the level of knowledge held by a sample that was typical of the total population. Despite most participants showing signs of competence, the results indicated a large gap. What this means is that there has been broad success in raising awareness about cybersecurity. However, these initiatives have clearly failed because a sizable segment of the population remains uneducated. We used SMOTE, or the Synthetic Minority Oversampling Technique, to balance out the distribution of data among respondents. Many distinct categorization models were created, including LR, DT, SVC, RF, and K-Nearest Neighbors (KNN), among others. Ensemble models that combined KNearest Neighbors (KNN) with Decision Trees (DT) or Random Forest (RF) performed superiorly than the separate classifiers. The research found that using SMOTE improved the accuracy of the ensemble model, which shows how effective ensemble approaches are and how important it is to fix class imbalance in predictive modeling.