News-popularity-prediction
The project aims to develop an effective learning algorithm to predict how popular an online article (news or story) would be before its publication by analyzing several statistic characteristics extracted from it.
The data
- The data is taken from here
Project objective
Classify articles in different classes based on how many shares (how popular) they can get.
Types of classification
- 2 class classification(High, Low)
- 3 class classification(High, Moderate, Low)
Algorithms used
- Residual Sum of Squares (RSS)
- BIC (Bayesian Information Criterion)
- Kernel SVM
- Random forest
other contributors
- Ashutosh Ranjan
- Ayush Joshi