Introducing Kyso 2.0

#Introducing-Kyso-2.0

December 4th, 2018

Collaborative and Reproducible Data Science

#Collaborative-and-Reproducible-Data-Science

The team have been working hard over the last few months to bring you the latest version of Kyso. As a lot of you already know, we've recently made significant changes to our cloud infrastrucuture, ensuring high-end performance for our users when operating in their workspaces. But we've also completely revamped our UI to make navigation more efficient, and released a new Jupyterlab extension that allows users to publish to Kyso's platform from anywhere with internet connection and a Kyso account.

These new functionalities bring us ever-closer to the team's vision of the future of data science. To that end, we're very excited to announce this latest release!

Notebooks and Workspaces

#Notebooks-and-Workspaces

The first difference users will notice when they log in is Kyso's new navbar. Click on the

+ New
button to either upload a notebook or to start a new workspace (Jupyterlab environment). From your avatar in the far-right corner you can also navigate to the blog & documentation pages, as well as to your own personal profile.

On the upload page, you can upload multiple files at once, but must choose a main notebook (.ipynb) file for Kyso to render on our frontend. Any other selected files will be attached and available if you open the publication in a workspace.

Personal Dashboard

#Personal-Dashboard

Clicking on My Dashboard will take you to the screen seen below, which is a list of all your existing workspaces and published notebooks. As you can see you now have the ability to start and stop containers as your require them. Clicking on the workspace link itself without first pressing Start will also spin up Jupyerlab, to which you'll be redirected when it's ready.

Loading output library...

Profile

#Profile

On your profile page you can edit your avatar and bio. This is a simple dashboard of all your published notebooks, the page that other users will see when they search for your work.

Loading output library...

We are continuing to improve the overall design of the notebook dashboards - & you'll soon be able to manually set a title and notebook description on this page.

Published Notebooks

#Published-Notebooks

Users now have the option to tag their published notebooks with desriptive terms. Taking the example below, my publication is an intro to the work of renowned economist, Thomas Pikkety, and I created the plots using the bokeh plotting library. Accordingly, I've tagged the notebook with pikkety, economics, and bokeh. Once any given tag achieves a pre-defined number of tags, they will appear as collections on Kyso's explore page.

Loading output library...

Here are some collections already on our landing page!!!

Loading output library...

We have rearranged the notebook's navbar too. Clicking on the comments icon will auto-scroll to the comments section at the notebook's end. Open in workspace will fire up a new Jupyterlab environment containing the notebook and any attached files. Click fork when reading another user's notebook to clone that study onto your own profile. You can then opt to open it in a workspace.

Note that you now have the ability to apply a default code visibility setting - perfect for educational notebooks, guides or those written for techincal readers. Select the Code hidden button on the right; next to each code visibility setting on the dropdown menu, you'll now see a Make default option.

Loading output library...

Clicking on the 3 dots on the right-hand side will open up the drop-down menu as seen above, where you can print, view previous versions, embed, share or download the publication.

Explore

#Explore

Search Kyso's explore page for interesting notebooks to read and perhaps fork. Notebooks on this page are ranked by publication date and popularity.

Loading output library...

Kyso's Jupyterlab Extension

#Kyso's-Jupyterlab-Extension

We're huge fans of Project Jupyter at Kyso, and we're going all-in on JupyterLab, the next-gen evolution of the Jupyter Notebook. JupyterLab is a fully extensible interactive computing environment, with all sorts of powerful features in a single customizable UI, file browser with rich outputs for images, CSVs, TSVs, and full terminal access.

Perhaps the feature we are most proud of is our new kyso publish extension. Pre-installed on Kyso's workspaces, this plugin allows anyone to publish to our platform from any Jupyterlab environment. We have made a bunch of example guides on how one can set up Jupyterlab environments on AWS, GCP, Azure and DigitalOcean virtual machines. This is of particular use to those, for example, running compute-intensive notebooks on a GPU-powered instance on any one of the above cloud solutions, that may not run as smoothly our machines. Now you can still publish your results as awesome blogs to Kyso.

Please find the install instructions on Github here.

Loading output library...

Upcoming Features

#Upcoming-Features

While we are very happy with the recent improvements made to the platform, we are looking to add to them very quick. Over the coming weeks, expect to see the following:

1. Better Github Integration

We removed the old Github feature with plans to implement a better app in the very near future. Soon, users will be able to import their Github repositories into Kyso, make changes locally & push to Github, which will be automatically reflected in your published notebooks on Kyso.

2. File Browser

At the moment, when you publish a notebook, all associated files and other notebooks in the same directory are attached. Soon, however, you'll be able to actually browse those files and even switch the notebook rendered on our frontend - meaning you can publish an entire multi-notebook repository as one publication.

3. Shareable Dashboards

As of now users can publish static blog posts (with interactive javascript-based plots) - but soon users will also be able to share live dashboards with python callbacks. This means you can create completley interactive python-based web apps.

4. Pro and Team Plans

Two new paid features will soon be avaible for those looking for more larger machines, access to private repositories, as well as the ability to share within teams, including notebooksand dashboards.

Note on Kyso's Free Tier

#Note-on-Kyso's-Free-Tier

As mentioned above, we will be implementing paid plans very soon. As such, it is important to highlight the features everyone has access to on the basic plan:

  • Unlimited public publications.
  • Run only 1 workspace at any given time. Users can have an unlimited number of workspaces.
  • Maximum of 3 hours before workspace is shut down at any one time.
  • Low CPU priority.
  • Number of total hours across all workspaces is limited to a maximum of 10 hours per month.

Please contact support@kyso.io for more info on upgrading to our pro and team plans, coming soon!