How to Install Python for Machine Learning or Deep Learning: A Comprehensive Guide

The installation of Python for machine learning or deep learning doesn't have to be a complex process. Here, we'll guide you through the steps for setting up Python, especially focusing on Python 3.6, which is currently well-supported with a wide range of libraries.

Step 1: Install Python 3.6

If you're newer to the Python ecosystem, a good starting point is to install Python 3.6. This version is currently well-supported, and many libraries are compatible with it. On some operating systems, such as Ubuntu, Python 3.6 comes pre-installed.

Step 2: Use Virtual Environments (Optional but Recommended)

As your projects grow and more complex dependencies are introduced, consider using virtual environments to containerize your projects. This approach makes it easy to manage different project dependencies. For example, you can use the pip package manager to install libraries like numpy or TensorFlow without affecting other projects.

Step 3: Consider Google Colab (Optional)

If you find that your local machine is limited in resources, especially when it comes to CPU or GPU, consider using Google Colab. It provides a free quota of CPU/GPU resources and a preset environment. Moreover, it offers a supportive community that can assist you in various machine learning and deep learning tasks.

Alternative: Anaconda Distribution

Your question implies that Python might not come built-in with your operating system. For this scenario, we will focus on the Anaconda distribution, which is a great alternative.

What is Anaconda?

The Anaconda distribution is a free and open-source package and distribution for data science and scientific computing on the Python and R programming languages. It includes Python (and R, but here we will focus on Python), scientific libraries (many of which are in the Python ecosystem), and the Jupyter notebook via the included Anaconda Navigator interface.

How to Install Anaconda

To install Anaconda, follow these steps:

Visit the Anaconda Python/R Distribution website. Click on the Download button. Select your operating system from the top bar. Choose the desired Python version and processor architecture that match your system. Install following the prompts provided by the installer.

The Anaconda distribution comes pre-installed with many popular scientific libraries and packages, which are optimized for your processor and architecture. For instance, it includes Intel's MKL DAAL and NVIDIA's CUDA if applicable, which can significantly speed up your machine learning and deep learning tasks.

Benefits of Using Anaconda

One of the benefits of using the Anaconda distribution is its extensive support documentation and compatibility with other Python environments. If you are familiar with Python in other environments, such as Linux, you will find little to no surprises. The Anaconda Distribution can coexist with other Python environments, even those built into most Linux distributions.

My personal experience as an experienced user of Linux and Python from the late 90s onwards has shown that the Anaconda Distribution can be used with no conflicts with the native Python environment or the need to manage separate Python modules in the OS environment.

Portability

An additional advantage of using the Anaconda Distribution is the portability of your projects. Once set up, your projects can run on any supported OS or CPU environment with minimal changes. This includes various cloud services, making it an ideal choice for deployment and experimentation.

Good luck with your machine learning and deep learning projects!