These instructions will guide you to discovering reliable (causal) rules from observational data.

=================
Preparing the executable
=================
------------
Option 1
------------
- Checkout the development branch of the realkd repository from https://bitbucket.org/realKD/realkd/overview.
- Prepare the jar files using maven from inside the root folder.
- The compiled SNAPSHOT jar will be in the "target" directory.

------------
Option 2
------------
- Use the jar file provided in this directory.

=============
Preparing datasets
=============
The dataset must be in one of the following file formats.
1. arff
2. xarf (https://bitbucket.org/realKD/realkd/wiki/model/data/xarf)

========================
Discovering reliable (causal) rules
========================
To discover reliable (causal) rules, we have to provide a job description file in the json format. A sample job file is provided in this directory. In short, we provide information about the dataset(s) in the "workspaces" field, and the computations to carry out on those workspaces in the "computations" field. For the detail information about the "computations" field, please refer to the "subgroupDiscovery.html" file inside the "kdondoc" directory.

To run the job, we simply provide the job description file as an argument in the command line.
	$ java -jar realkd-0.7.1-SNAPSHOT-jar-with-dependencies.jar sgd.json
The result of the computation will be stored in the "output" directory.

=========
Contributors
=========
Kailash Budhathoki (kbudhath@mpi-inf.mpg.de)
Mario Boley (mario.boley@monash.edu)
