This is a small project trying to predict the user reputations from Stackoverflow with regression models. It uses features derived from graph mining techniques, particularly graph centrality measures.
This project depends on graph-tool ([https://graph-tool.skewed.de/]) and keras.
- Download Stackexchange dumps from [https://archive.org/details/stackexchange]
- Create graphs runnning
extract.pyfrom inside the Stack Exchange folder - Create graph features running
centrality.py - Running
analyse.pygiven a features file will create various singlevariate regressions as well asa multivariate regression and a neural network regression. It also tries to predict other networks with the learned regression model