Cloud-Based Notebooks

Cloud computing, which was a buzzword a decade ago, is now the lifeline for many organizations. The cloud has become immensely popular as it is not only convenient but also cost-effective.

When it comes to Data Analytics, the cloud sentiment is no different; in fact, it is often more convenient than traditional in-house solutions migrated.

Today, there are many cloud-based Jupyter Notebooks. Using these systems allows you to skip the Anaconda installation step altogether and start your analytics immediately.

Google Colab

Google provides a free cloud notebook platform called Colab. You can access it at: https://colab.research.google.com

Salient Features
  • Pre-installed Modules: Most modules in Anaconda are available, along with many others.
  • Temporary Environment: While you can install additional modules, the environment is temporary; any custom installations must be reinstalled when a new session starts.
  • Jupyter-Based: It is built on the popular Jupyter Notebook interface.
  • Storage: You can save .ipynb files to your Google Drive (free up to 15GB). A Google account is required to save and run notebooks.
  • Collaboration: Just like Google Docs, you can collaborate with others to edit or share notebooks with specific permissions.
  • Data Persistence: Files uploaded directly to the temporary environment are deleted after the session. To store datasets permanently, use persistent services like Google Drive or Google Cloud Storage.
  • Performance: The environment provides approximately 13 GB of RAM and offers free access to GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units) for accelerated performance.
  • GitHub Integration: You can open notebooks directly from GitHub, save changes to a repository, or work with private repositories.

Note that when a Colab notebook is shared with you, you must copy the file to your own Drive before you can run it. While computing power is currently free, this may change in the future.

Troubleshooting & Tips
  • Restarting: If results become unpredictable or you lose track of active variables, you can clear the environment by selecting Runtime -> Restart runtime (in Colab) or Kernel -> Restart (in Jupyter).
  • Persistent Issues: If restarting doesn't work, try shutting down the kernel, closing the file, and reopening it for a completely fresh start.

Google colab

Binder (MyBinder)

The MyBinder project (accessible via Rackspace and other providers) provides a free Jupyter Notebook environment as well, at: https://mybinder.org/v2/gh/jupyterlab/jupyterlab-demo/master

Note: The system recycles your instance after 10 minutes of inactivity. To avoid losing your work, download your notebook files frequently. You can start a new notebook by choosing New -> Python 3, as shown in the image below.

Rackspace