A prediction system that can automatically guess the star ratings trained from 100,000 reviews on Amazon.com.
Course: CIS 520, Machine Learning, Fall 2011, University of Pennsylvania
Teamwork: Yayang Tian, Tao Feng, Wenbin Zhao
Skills: Matlab, Python, machine learning
- Increased accuracy from 40.1% to 81.3%, and decreased RMSE from 1.460 to 0.853.
- Implemented various machine learning methods, including feature selection like PCA, stemming, metadata, part of speech, and information gain, as well as mathematical models like Naive Bayes, Ada-boost, Logistic Regression, SVM, Intersection Kernels, and EMs.
- Got TOP one performance for a long period in class and awarded by Prof. Ben.Taskar.