Experimental-HazardRegression (Attributing Hacks) with Trend Filtering

Features

• Allow dynamic parameter spaces
• No features preprocessing
• Online learning with isotonic fussed lasso signal approximator

Data

• Currrently support libsvm format

(Note: under developing)

Usage

• git clone --recursive git@github.com:ziqilau/Experimental-HazardRegression.git
• cd Experimental-HazardRegression; git checkout 9d1358e; make -j8;
• cd example; ipython notebook monotone_demo.ipynb