Setting up Python virtual environment

Exploring fields like Machine Learning, Data Science or Natural Language Processing becomes quite easy with the use of Python, Matlab (Octave) and R.

When we consider python, the most common problem encountered when dealing with various packages is that, due to continuous development, a particular version of a package becomes outdated pretty quickly and then you have to deal with updating them, which might break some existing functionality in your projects, and so on.

To avoid such problems, there is something known as Virtual Environment which creates isolated boxes of python environment in which you can have python packages as per your needs.

Following are the steps that you need to follow to set up a virtual environment in Ubuntu, with Python 2.7.

Type pip -v in your terminal, if package not found error is displayed, install it as described in next step, else skip the next step.
Install python package manager pip using sudo apt install python-pip if not already present. (For python3 it will be a different command)
Install Virtual Environment using sudo pip install virtualenv. Note whenever you install some python package with elevated privileges (sudo) it will be installed globally.
Go to a directory where you store your codes cd /your/code/directory.
Make a new virtual environment using virtualenv <name_of_virtualenv>. Lets call it venv then, virtualenv venv,will create a new virtual environment named venv in which you can install python packages as per your project requirements. It is recommended that you have separate virtualenv for ech of your projects.
Activate the venv using source venv/bin/activate, and later you can deactivate it using deactivate. Note that source command needs an exact path of your virtual environment, meaning, if you are in some different directory, change the path in the command appropriately. deactivate can be used from anywhere.
Start installing your project specific packages using pip install <package_name>. Lets install nltk, using, pip install nltk.
Also you can upgrade packages with pip install --upgrade <package_name>.
You should never push your virtual environment while collaborating with other users on GitHub or BitBucket (add , in our case `venv`, to `.gitignore` to tell git to ignore it) since otherwise it would result in lots of conflicts. But then how will your collaborators know of any package that you have installed? See next step.
For that, you must use requirements.txt file. Using which each collaborator can install packages in their own separate virtualenvs.
Using pip freeze command you can get a list of packages that are installed in your system. If you run this command after activating virtual environment, this will give packages that are installed in a your current virtual environment.
To make a requirements.txt file for a specific virtual environment use, pip freeze > requirements.txt after activating it.
You can push requirements.txt file, and your collaborator can use pip install -r requirements.txt, after activating his/her virtual environment which will install all the packages that are listed in requirements.txt.

Comment below for hugs and bugs.