This repo is used as a collaboration platform for the Kaggle competition Toxic Comment Classification Challenge. The programming language for the repo is Python.
If developing on windows, Git bash is recommended. It includes a small Linux-like environment based on MSYS2 (which in turn is based on Cygwin.)
Python requires an IDE or text editor of choice e.g.:
- PyCharm (https://www.jetbrains.com/pycharm/)
- Sublime Text (https://www.sublimetext.com/)
Install Python 3.5.X or 3.6.X as Tensorflow requires it.
For Windows:
• Download Windows x86-64 executable installer from [python.org](https://www.python.org/downloads/)
• Tick the “Add Python to PATH” option
• Click “Custom Installation” and leave the first page unchanged.
• Choose to install python to the directory “C:/Python3X” where X is the Python version preferred.
• Click “Install”
• Check if the installation was successful by opening the command prompt and type: python3 and verify that a similar response is obtained as below:
For OSX:
Follow: https://medium.com/@GalarnykMichael/install-python-on-mac-anaconda-ccd9f2014072
You also need to install virtualenv
, as that is might not bet installed by
default:
pip install virtualenv
If you do it from a VCC computer, at VCC, you will likely have a problem with the VCC proxy. See below for how to work around that.
Before starting development a so-called virtual environment is good to setup. Do this by executing
setup/setup_virtualenv.sh
This will install some nice-to-have python modules like Keras, Tensorflow and Matplotlib.
Now do the steps below under Every time initialization
To activate the local python environment, go to the root of the Git repo and execute
on Linux: source env/bin/activate
on Windows: source env/Scripts/activate
Your prompt should now show (env)
as the first characters to the left.
In order to install new python packages, run
pip install <package>
If you do it from a VCC computer, at VCC, you will likely have a problem with the VCC proxy. See below for how to work around that.
If you want others to use the add-on, add it to the list of modules to install
by editing the file setup/REQUIREMENTS.txt
and commit it.
When working from inside the VCC network a proxy is placed between the VCC network and the rest of the internet. We need to tell the tools to use this proxy, and not attempting to connect to the internet directly. In a terminal run:
export http_proxy=proxy.volvocars.net:83 https_proxy=$http_proxy
and then retry your previous action.
Files can be manually uploaded to the competition site, or the submission script can be run:
./submit.sh <file_to_submit>
This script will ask for a breif description of the submission and Kaggle username and password.
Don't push competition data to repo.
Slack-workspace: "Kaggle-ED"