Tagged assignments

Guest Post: Docker And the Data Scientist

Educational institutes and educational professionals often face a problem when it comes to creating a common platform where educators and students can view and share code. One such university in Turkey had to find a way to address a common complaint from students viz. their compute environments were different when compared to the testing machine.

 

The professor at Bilkent University in Ankara, Turkey, decided to use a technology called Docker to power a web-platform that can create lab instances and grade assignments.

 

So, what is Docker? We will answer that in a while. But before Docker was available, the next best solution was to use virtual machines. However, these machines needed to be extremely powerful and consequently required an expensive infrastructure, which most institutes couldn’t set aside a budget for. Students were forced to log on to a shared server which inadvertently negatively affected each other’s programs, or worse, crashed the whole infrastructure. Needless to say, it was impracticable to assign a virtual machine to each student.

 

They used Docker to build a web-based application called Programming Assignment Grading System (PAGS). A similar technique can be of adopted by universities for creating lab instances and grading assignments for data science classes.

 

Although we haven’t formally defined what Docker is, the above example demonstrates what Docker can do. The rest of the article focuses on Docker and how it can transform the education and data science industry.

 

The article is divided into 4 sections. First, we’ll start with an introduction on Docker and Docker containers. Then, we’ll answer the question, “Who is Docker for?” The third part will give you an overview of how Docker is a useful tool for data scientists. In the final section, we’ll dive into a couple of  interesting use cases for Docker in data science. Let’s get started!

What is Docker?

 

Docker is the leading software containerization platform that is being actively developed by Docker Inc. It is an open source project that is designed to help you create, run and deploy applications inside containers.

 

So, what is a container? A container, by definition, comprises all the dependencies, libraries and other related files required to run an application. Once you’ve created a container for your application, you can run it on any Linux machine regardless of the way your underlying machine is configured. If the machine that you’re using at one end is Ubuntu, and it’s Red Hat at the other end, fret not! Docker is precisely meant for situations like these.

 

You can create a snapshot of a container and this snapshot is generally known as an image. Conversely, you can call a container an instance of a docker image. Docker images are inert and immutable files. When someone asked the difference between an image and a container on StackOverflow, a web developer named Julian came up a quick analogy. “The image is the recipe, the container is the cake”, he said and that just sums it up.

 

You can store Docker images in a cloud registry like Docker Hub. There are numerous user-contributed Docker images that should cover almost all the general use cases. You can also create and share your private Docker images with your co-workers and your organization. Alternatively, you can push them into a public repository so as to return it back to the community.

 

The concept of Docker is very similar to that of a Virtual Machine (VM). However, virtual machines are very demanding beasts and run considerably slower on a less powerful hardware. A VM works in such a way that it allows a piece of hardware to be shared between VMs. This allows you to run one or more virtual operating systems inside your host operating system. But you might need to upgrade your processor if you’re seriously planning to run your software on a virtual machine.

 

Unlike a VM, Docker uses the host kernel instead of creating new kernel instances. The virtualization happens at the kernel level and not at the topmost level. The Docker encapsulates everything that’s required for running the application on that host machine. This tremendously improves the performance of the application and reduces its size. What gives Docker the significant lead is the fact that it enables separation of concerns between the infrastructure, IT Ops, the developer and the application. This creates a positive environment for enhanced innovation and collaboration.

Who is Docker for?

Docker is essentially a container platform largely aimed at businesses. It enables IT businesses to efficiently select and administer a complete application, without the fear of an infrastructure or architecture lock-in.

 

Enterprises use Docker for everything from setting up their development environment to deploying their application for production and testing. When you need to build more advanced systems, like a data warehouse comprise of multiple modules, containers make a lot of sense. You can actually save several days of work that you’d otherwise have to spend configuring each machine.

 

However, the Docker platform isn’t just relevant to developers and enterprises alone. It’s actually a pretty useful tool for data scientists, analysts and even for schools and colleges. There are educational institutions and universities that are keen to transform digitally but are held back by their existing infrastructure.

Docker and Data Science

Why should you use Docker if you’re a data scientist? Here are three reasons pointed out by Hamel Hussain over at Towards Data Science:

Reproducibility

If you are a professional data scientist, it is imperative that your work can be reproduced. Reproducibility helps facilitate review by your peers, ensure the analysis, model and application that you have built can run unhindered which makes your deliverables both robust and time-tested.

 

As an example, let us assume that you have built a Python model, however, it has not proven to be enough to run pip-freeze and transfer the resulting file to a colleague. This would largely be because of Python-specific dependencies.

 

Imagine if you could find a way around manually moving the Python dependencies like  the compiler, config. files, drivers, etc. You can be free of Python-related dependencies by simply bundling everything within a Docker container. This not only reduces the task of others having to recreate your environment, it also ensures that your work is much more accessible.

Ability to Port Your Compute Environment

If you are a data scientist who is specializing in Machine Learning, the ability to frequently and efficiently change your computing environment has a considerable effect on your productivity.

 

It is often the case that the work of data science starts with prototyping, research, and exploration. This doesn’t essentially need special computing power to start. That said, often comes a stage where multiple compute resources can prove quite helpful in increasing the speed of your workflow.

 

A number of data scientists find themselves limited to a local computing environment largely because of a perceived hindrance of re-creating their individual local environment onto a device remotely. Here, Docker makes the difference. It allows you to port your work environment, including libraries, files etc in just a few clicks. Additionally, the ability to swiftly port your computing environment is a substantial advantage in Kaggle competitions.

Enhance your Engineering Skills

Once you are comfortable with using Docker, you can then deploy models as containers that can help make your work readily accessible to other users. Additionally, various other applications that you may require as part of your data science workflow interaction may already exist in a container within a Docker application.

Use Cases for Docker in Data Science

By making your applications portable, cheaper and more secure, Docker helps to free up time as well as resources that can be spent on other important things. It can help transform IT without the need to re-tool, re-educate or re-code any of your existing applications, staff or policies.

 

Here are just a few of the use cases of how Docker can help different organizations:

Docker for Education

Let’s revisit the Docker use case that we discussed in the introduction. The faculty at the university used Docker to create a container to run their application called PAGS. It allowed the students to have the same environment for their compute machines and test machines without the need of a VM.

 

Docker provides a common environment that can run on a container on any given Linux machine. This almost always guarantees it to run with similar results on a different machine using the same container. Without Docker, this would have required more infrastructure and resources that they didn’t have.

 

Another particularly interesting scenario is setting up lab instances. Dependending on how you want a machine to be configured, you can take a snapshot of it to build a Docker image. You can then pull the snapshot into all other lab instances saving you time and resources.

Docker for Data Science Environment Set Up

Consider a scenario where you need to explore a few data science libraries in Python or R, but;

  1.  without spending a lot of time installing either language on your machine,
  2.  browsing and figuring out which dependencies are essential and
  3.  finally getting down to identifying what works best for your version of Windows/OSX/Linux.

 

This is where Docker can help.

 

Using Docker, you can get a Jupyter ‘Data Science’ stack installed and ready to execute in no time flat. It allows you to run a ‘plug and play’ version of a Jupyter data science stack from within a container.

 

To start, you would need to first install Docker Community Edition on your machine. Once done, restart your machine and get your Jupyter container set up. Prior to running a container, you would need to specify the base image for the container

 

In most cases, the image that you’re looking for has already been built by a prior user and includes everything needed to use a fully loaded data science stack of your choice. All that needs to be done is to specify a pre-defined image that Docker can use to start a container.

Conclusion

In this article, we have just hit the top of the iceberg in terms of what can be done with Docker. We have focused only on specific areas of Docker that a data scientist may most often encounter. Below are some further sources that can help you during your journey of using and implementing Docker.

 

  1. Basic Docker Terminologies
  2. Useful Docker Commands
  3. Dockerfile Reference
  4. Pushing and Pulling to and from Docker Hub

This was a guest post by: Limor Wainstein

Limor is a technical writer and editor with over 10 years’ experience writing technical articles and documentation for various audiences, including technical on-site content, software documentation, and dev guides. She holds a BA in Sociology and Literature and is an MA student in Science, Technology, Society (STS) at Bar-Ilan University. Limor is focusing her studies on the sociology of technology and is planning her research around coworking spaces in Israel.

Intro to DS Assignment Sites

As an instructor, I want to provide high-quality assignments that are focused (so they achieve the learning objective), engaging (so they aren’t bored), and well supported (so they don’t end up frustrated). In an ideal world, I’d have time to write, test, debug, and administer all my own, course-tailored assignments that meet these goals. I, however, do not live in an ideal world, nor have enough graduate/undergraduate minions to mimic this ideal world. Instead, I’ve turned to using a few sites that already host assignments, resources, and even include auto-grading (without me needing to learn/setup the system).

Learn2Mine (L2M) is the first site I used in conjunction with my Data Mining course, and more recently my Introduction to Data Science course. Learn2Mine is a free, open source platform developed at the College of Charleston (CoC). While I have only really made use of the contents already there and CoC’s hosted site, you can contribute, or host your own version by getting the source directly from github. Dr. Anderson is fairly responsive about keeping the site running and grading.

The positive features for L2M (beyond being totally free/open source) are that it includes a mix of both introductory programming assignments and several more advanced machine learning/data mining lessons. It even has several search algorithm lessons (which I tend not to use). All of the lessons include auto-graded response boxes which also provide limited feedback of the errors generated when comparing submitted work to answers. There is also an interface for instructors to create their own ‘courses’ which consist of a series of the lessons on L2M. This allows the instructor to see student progress through lessons and download a grade-book in spreadsheet format.

Downsides for L2M are in-line with what you pay for (or invest in time-wise). Even though there is feedback when students get answers wrong, this often just consists of the identification of mismatched output lines (so pretty sparse). Students often get very frustrated trying to figure out what they are missing. This is exacerbated by the fact that often the instructions are unclear or insufficient to allow students to simply do the lessons. Also, as might be expected from a locally built/maintained project, there are a lot of “polish” features missing, such as being able to reorder assignments in a course, or associate a name with an account. Students have an account associated with the email they login with so it can sometimes be challenging to connect records with students. Overall, I’ve been considering phasing L2M out of my normal assignment structure, though the possibility of hosting my own local version and implementing different, more explained lessons has also been tempting.

The prime contender to replace L2M for me has been DataCamp. I’ve know about DataCamp for a while now but had the first chance to actually use it and make assignments from it this spring when I was looking for data visualization lessons (see visualization resources post). I’ve gone through a few lessons myself and found DataCamp to basically be exactly what I’d want/envision online course-work to be. Most courses consist of short videos (a best practice) followed by several guided coding exercises. DataCamp is not (sort of) free, which turns out to be a pro and a con.

If it’s not free, why is DataCamp going to replace L2M for me? Great question. Because, for academic purposes, Datacamp IS free. If you are an instructor for an academic institution teaching a course with 10+ students in, you can request free, premium access for students enrolled in your course(s). That access is limited (they give you 6 months), but hey, it’s free. What else makes DataCamp a nicer replacement? First the coding exercises are scaffolded, that is, early exercises have more prewritten code while later exercises require you to remember and use what you’ve already learned. In addition, the coding exercises have reasonably helpful error messages and help often allowing you to more accurately debug code. They’ve also got built in hints/help available, so a student can’t get permanently stuck. Using those however decreases the “exp” they gain, so you can still track how success a student has been without help. The other major advantage is that DataCamp has a SIGNIFICANTLY larger set of lessons/courses available to pull from.

There is no free lunch in data/computer science though. DataCamp does have a few downsides. Perhaps the biggest is the granularity available in assignments. You have three choices, “collect xp”, “complete chapter”, or “complete course”. Given that a chapter is really the smallest cohesive learning unit on DataCamp, this makes a lot of sense educationally. However, that also means it’s not exactly an alternative for giving individual lab/homework assignments. Instead, it would serve best as a resource/major assignment related to learning how to program in python/r, or a bigger topic.

Finally, I want to mention Gradescope. Gradescope isn’t data science educational site. Instead it’s a jack-of-all trades which can help ease the burden of assignments and grading. If DataCamp took L2M and removed granularity/options, Gradescope (in this context) goes the other direction. Lots of faculty use it for all kinds of courses, from computer science or mathematics to writing. Given its purpose, Gradescope doesn’t have any specific assignments (maybe that was obvious). Instead, it can serve as an autograder or collection site for your assignments. I’ve included it here for those that might already have assignments (or who get them from others) but still want a speedy, simple way to get feedback to students.

I’d be remiss if I didn’t point out that there are some alternatives to DataCamp, depending on your goals. If all you need students to do is learn to program (not necessarily in a data-centric style) try Codecademy or explore Code.org. I also know there is an alternative to Gradescope (but I couldn’t track down the name/site if someone knows, please email me or leave a comment). What I recall is that the alternative is NOT free, but does provide better support and scaling. You might also consider what options are available or integratable with your learning management system (DataCamp IS…but maybe not by you..).

Hopefully you found this post informative, if you’ve got other suggestions of websites with assignments (particularly data-science related) please let me know or leave a comment.