- Evaluated student performance predictive model using artificial neural networks, naïve bayesian and decision tree
- Used ensemble methods of bagging and boosting, random forest to improve the model accuracy after cleaning and processing student data collected from Kaggle for null/missing value removal/imputation, dimension reduction and outlier detection
- Attained an accuracy of 92.34% after testing the model. Depicted visualization of explorative analysis via Python libraries on Jupyter identifying key factors affecting student performance
- Platform : Jupyter Notebook
- Programming Languages : Python
- Python Libraries : pandas, numpy, seaborn, matplotlib, sklearn
Gender Histogram
Subject Topic Histogram
Student Grade Distribution
Student Class Section Distribution
Student Nationatility Distribution
Student Involvement in Online Education Board Histogram
Parent School Satisfaction of different Grades
Student distribution of those who Raise Hands in class based on their Nationality
Nationality vs Gender Distribution
Scaled Algorithm Comparison
Scaled Ensemble Algorithm Comparison