This is a class project for stat 535 in Unversity of Washington. The project link is http://www.stat.washington.edu/courses/stat535/fall14/project.html
Duty Description: Binary classification (buzz/no buzz) on Twitter
Dataset:
- Training: 100,000 samples, 77 predictors, 1 response
- Testing:
- http://ama.liglab.fr/resourcestools/datasets/buzz-prediction-in-social-media/
Two supervised methods are implemented:
- RBF-SVM: SVM with Gaussian Kernel
- SVM with other kernel (pending)
SVM Packages:
- SVM-torch: C++ Package
- SVM-light: C Source
- LibSVM: C++ and Java Source, R interface package, e1071. http://www.csie.ntu.edu.tw/~cjlin/libsvm/