Installation

Requirements

• Python 2
• IPython

Setup

Linux

IPython

If you have pip you can install IPython with it:

sudo pip2 install ipython

parallel_leaf

If you have git you can obtain the source code by:

git clone https://github.com/wasit7/parallel_forest.git

Or you can download a source code from the repository page

Run

Linux

If you intend to run this project on local computer, please make sure that engines' working directory is the project directory.

In order to have project's directory as working directory of the engines, add this line to code in scmaster.py before any remote import on engines.

self.dview.execute('os.chdir("path_to_your_project_directory")')

then start ipcluster by execute this command:

ipcluster2 start --n=number_of_engines_you_want

OR just run the command above in the project directory.

Obtain the sample dataset by execuete:

python2 dataset_pickle.py

This will create the sample dataset file to the /traning. You have to rename it to datasetXX.pic, where [XX] is index of an engine you have to change the index according to the number of engine you have. For example, a number of engines is 8 the number [XX] are 00, 01, 02, ..., 07.

Then you can start the training process by execute:

python2 scmain.py

The parallel_leaf will use the dataset files in the /training for training process. Make sure that you have enough dataset files for your engines, and it was named properly.