Heteroscedastic Discriminant Analysis (HDA) is a machine learning algorithm for dimensionality reduction. It is implemented as a package for the open source program WEKA.
#Building Ensure you have the following programs and tools installed on your computer before proceeding.
- Java's JDK
- apache-ant
- development version of Weka (Tested with version 3.7.12)
- git
The next step is to ensure that weka.jar is in the $CLASSPATH variable. On Arch Linux if weka was installed from the AUR, the jar is found in /usr/share/java/weka/weka.jar. To add it to the CLASSPATH on Arch add this line to your ~/.bashrc and then update your systems class path we source the bashrc.
export CLASSPATH=/usr/share/java/weka/weka.jar:$CLASSPATH
source ~/.bashrc
Next step is clone the repo,
git clone https://github.com/StephenNu/HDA.git
After that move into the location of the repo and make the package by the following commands
cd HDA
./build.bash make_package
Now in weka open the package manager and click "File/URL" under the unoffical section. Navigate to where the repo was cloned to, and go into the dist/ folder, and select HDA.zip Click okay, restart weka and HDA will be installed under filters.
#Tests Once HDA is installed, you can run the HDA tests by going to the weka command line, and running the following command
java weka.filters.HDATest