MS Project Defense - Rong Qi

A New Ensemble Machine Learning Algorithm

Rong Qi, MS Student

Friday, November 17, 2017, 1:00 pm
Dunn Hall 311

Committee Members:
Prof. Vinhthuy Phan, Chair
Prof. Nirman Kumar
Prof. Deepak Venugopal

Supervised learning is the machine learning task of inferring a function from labeled training data. There are several supervised classification machine learning algorithms. The choice of which specific learning algorithm we should use is a critical step. In this study, a new ensemble supervised classification machine learning algorithm is created to get higher accuracy than each individual algorithm. This algorithm combines 5 common supervised classification machine learning algorithms (Decision tree, logistic regression, naive bayes, SVM, random forest). For each algorithm, grid search was used for hyperparameter optimization. The best parameters were generated after training. The model was saved and used for testing. The top-3 accuracy algorithms were selected and I used voting method to combine these 3 model predictions into ensemble predictions. This new classifier showed better accuracy in most testing cases across mutiple datasets when comparing with other algorithms and with the voting class implemented in scikit-learn.