Anaconda: Your Complete Toolkit for Machine Learning Experiments

Setting up Anaconda on Ubuntu for seamless ML experimentation

Introduction

Anaconda is a powerful open-source distribution of Python and R programming languages for data science and machine learning projects. It comes with a plethora of pre-installed libraries and tools that streamline the setup process, making it the preferred choice for many data scientists and researchers.

Installing Anaconda on Ubuntu

Follow these steps to install Anaconda on Ubuntu:

  1. Download the latest version of Anaconda from the official website: Anaconda Distribution.
    2. You can also dowload the older version from the Anaconda Archive.
  2. Open a terminal window.
  3. Navigate to the directory where the Anaconda installer script was downloaded.
  4. Run the following command to start the installation:
bash Anaconda3-2023.09-0-Linux-x86_64.sh

Follow the prompts on the screen to complete the installation. Make sure to add Anaconda to your PATH variable during the installation process.

Miniconda

Miniconda is a minimal installer for Anaconda. It includes only conda, Python, and other essential packages, allowing you to install additional packages as needed. Miniconda is a great choice if you want more control over the packages installed on your system.

You can download Miniconda from the official website and follow similar installation steps as Anaconda.

Creating a New Environment

Once Anaconda is installed, you can create a new environment for your machine learning experiment. Environments allow you to manage different sets of libraries and dependencies for different projects.

To create a new environment named "ml_env" with Python 3.8, run the following command:

conda create --name ml_env python=3.8

Activating the Environment

To activate the newly created environment, use the following command:

conda activate ml_env

Installing Packages

With the environment activated, you can install the required packages for your machine learning experiment. For example, to install TensorFlow, run:

conda install tensorflow

Similarly, you can install other packages such as scikit-learn, pandas, matplotlib, etc., using the conda install command.

Running a Machine Learning Experiment

Now that you have Anaconda set up and your environment configured, you're ready to run your machine learning experiment. Write your Python code using your preferred IDE or text editor, and execute it within the activated environment.

For example, if you have a Python script named experiment.py, you can run it using the following command:

python experiment.py

Deleting an Environment

If you no longer need an environment, you can delete it to free up space and keep your system clean. To delete the "ml_env" environment, run:

conda remove --name ml_env --all

Jupyter Notebook Server Management Guide

🔍 List Running Jupyter Notebook Servers

jupyter notebook list

This shows all running notebook servers along with access tokens and URLs.

Set or Change Jupyter Notebook Password

jupyter notebook password

This creates or updates your notebook login password.

Add Conda/Virtualenv Environment to Notebook Kernels

python -m ipykernel install --user --name myenv --display-name "Python (myenv)"
  • myenv: the environment name
  • Python (myenv): name shown in the Jupyter interface

Start Notebook on Custom IP and Port

jupyter notebook --ip=0.0.0.0 --port=8888 --no-browser

For persistent settings, update:


c.NotebookApp.ip = '0.0.0.0'
c.NotebookApp.port = 8888
c.NotebookApp.open_browser = False

🛠 Generate Jupyter Notebook Config

If not already present, generate it with:

jupyter notebook --generate-config

Enable Notebook Extensions (nbextensions)


pip install jupyter_contrib_nbextensions
jupyter contrib nbextension install --user
jupyter nbextension enable collapsible_headings/main
jupyter nbextension enable toc2/main

Enable other extensions through the Jupyter dashboard interface.

Use JupyterLab Instead of Classic Notebook


pip install jupyterlab
jupyter lab --ip=0.0.0.0 --port=8888 --no-browser

JupyterLab offers a modern UI with file browser, terminals, tabs, etc.

Enable HTTPS for Secure Access

Generate self-signed SSL certificate:


openssl req -x509 -nodes -days 365 -newkey rsa:2048 -keyout mykey.key -out mycert.pem

Update config file:


c.NotebookApp.certfile = '/path/to/mycert.pem'
c.NotebookApp.keyfile = '/path/to/mykey.key'
c.NotebookApp.ip = '0.0.0.0'
c.NotebookApp.port = 8888
c.NotebookApp.open_browser = False

Auto-Start Jupyter Using systemd

Create file: /etc/systemd/system/jupyter.service

[Unit]
Description=Jupyter Notebook

[Service]
Type=simple
PIDFile=/run/jupyter.pid
ExecStart=/usr/local/bin/jupyter-notebook --config=/home/youruser/.jupyter/jupyter_notebook_config.py
User=youruser
Group=youruser
WorkingDirectory=/home/youruser
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

Enable and start service:


sudo systemctl enable jupyter
sudo systemctl start jupyter

View Logs for Debugging


# For manual startup logs
journalctl -xe

# If running via systemd
sudo journalctl -u jupyter -f

Conclusion

Anaconda provides everything you need to kickstart your machine learning journey on Ubuntu. With its user-friendly package manager, environment management, and extensive library support, Anaconda simplifies the setup process and allows you to focus on building and experimenting with machine learning models.

Install Anaconda today and start exploring the exciting world of data science and machine learning!