University Recommender
- Scraped graduate profiles from Edulix using BeautifulSoup and predprocessed data.
- Implemented Wilcoxon Rank-Sum p-Test, Generative Additive Models, Kernel Density Estimation to study correlations between “desicion” variable and other variables(continuous)
- Trained and evaluated SVM, logistic Regression, XGBoost, RandomForest and AdaBoost.
- Attained the highest AUC-ROC(0.83) and AUC-PR(0.87) with XGBoost and chose it as our final model for admission prediction.
- Implemented KNN to recommend top ‘k’ universities that best suits with the profile.
- Created an interacive web page to deploy the models through flask.