Beyond Jupyter Notebook

Beyond Jupyter Notebook

For a lot of data scientists, notebooks are the only way they have learned to write code. Though Jupyter Notebook has well-known defaults, there is no better Python tool actually for exploratory analysis and quick tests. The point is that you are probably lacking something if most of your work relies on Jupyter Notebook.

For a lot of data scientists, notebooks are the only way they have learned to write code. It’s quick to set up, easy to run and we don’t have to really understand what’s under the hood.

Though Jupyter Notebook has well-known defaults, there is no better Python tool actually for exploratory analysis and quick tests.

Still, when facing more real-world issues, collaborating within a larger team or keep reproducibility, only using notebooks for every task is quite limiting.

The point is that you are probably lacking something if most of your work relies on Jupyter Notebook.

From experiment to production

There is a great similarity between data science stuff and the pharmaceutical industry.

Any drugs we can find on the market today have been discovered, studied by one or many scientists. They had to experiment with new chemical combinations to find the molecule that would fight the corresponding disease. The first step lives in their laboratory.

It is the same thing with data science: you are facing a problem and try different solutions in your notebook to find a correct solution.

When the pharmaceutical solution passes the laboratory test it moves to a whole other world before reaching out to your care shelf. To do mass production, pharmaceuticals engage a lot of assets in production factories, security, and all sanity checks needed.

Go back to our world, we can see these production assets as the beard guys: data engineer or software engineer who are technically closer to the real final product. They don’t use Jupyter Notebook because it is not designed for that kind of stuff and they have much more constraints than you with your notebook to fit the solution in a real-world application and infrastructure.

What is production?

We will go straight here while there will be a plain article to write about this.

Production is a state of development or a running application that reaches the real world, that engages people and their decisions. Therefore the underlying code and infrastructure that serve this environment have to be under perfect control to be sure that there is no downtime and no error for the end-user.

An end-user is either a real customer or people inside your company that drives their decisions based on your report, dashboard, or algorithm. In fact, it can be also you: think of a database that you are querying, if it is down for any reason you won’t be able to do your work.

To fulfill the requirements of such an environment, code has to be designed in a way that when an error occurs it must be easy to see and resolve. It should be also possible for anyone to upgrade the code without too many headaches — that way you’re not alone with your codebase and the company can still use what you produced when you leave.

To do so, it’s easier when your code is clean, throw good message errors, and informs when important steps are passed. Respecting coding guidelines, do code reviewing through merge requests, and have in mind some design patterns are often good things to enforce when you are dealing with production responsibilities.

All these points are far from our minds when we are writing into Jupyter Notebook. We generally write as we think, test, come back, re-test, etc… The code is often not clean, doesn’t deal with exceptions, nobody but you read it, etc… And this is perfectly normal, we are experiencing.

The Software Engineering shift

Most of the fancy machine learning and experimentation part is not that important if it doesn’t serve a business or an “applicable” goal.

Nowadays the technology we have to build great and powerful models don’t require a Ph.D.: almost every developer with some computer science background can use Scikit Learn for example.

It is the same at the other side of the spectrum: you don’t need to be a Linux master to understand serverless applications, containerization, and basic production requirements and assets.

When we go out of scholarship we often see the data stack split over three different positions: Data Engineer, Data Scientist, Data Analyst.

But what if we can gather all the requirements of these positions into one job?

This is how is emerging the “Full Stack Data Scientist” position. Though the word can be a bit overblown, there is in fact a lot of value in considering such a possibility.

In this kind of position, you would need to cover a lot of tools. You will have to learn a lot of things. But as data science becomes more mature now, we can find a bunch of blogposts, API, and high-level frameworks to easily bring our code closer to production and business requirements.

In fact, the position is already there and you probably already have the skillset to do so: there are not so many teams with big data stack relying on several data-scientists, data-engineers, and analysts. Most of the jobs are titled “Data Scientist” but are often roles that require many different skills and experiences.

Most people working with data have some computer science backgrounds. For companies, it would be much more valuable to allow people to extend their skills a little bit rather than fill an NLP researcher position with a guy who doesn’t know what is a server…

data-science jupyter-notebook python developer

Bootstrap 5 Complete Course with Examples

Bootstrap 5 Tutorial - Bootstrap 5 Crash Course for Beginners

Nest.JS Tutorial for Beginners

Hello Vue 3: A First Look at Vue 3 and the Composition API

Building a simple Applications with Vue 3

Deno Crash Course: Explore Deno and Create a full REST API with Deno

How to Build a Real-time Chat App with Deno and WebSockets

Convert HTML to Markdown Online

HTML entity encoder decoder Online

Why Jupyter Notebooks are the Future of Data Science

Why Jupyter Notebooks are the Future of Data Science. How Jupyter Notebooks played an important role in the incredible rise in popularity of Data Science and why they are its future.

Basic Data Types in Python | Python Web Development For Beginners

In the programming world, Data types play an important role. Each Variable is stored in different data types and responsible for various functions. Python had two different objects, and They are mutable and immutable objects.

Hire Python Developers

Are you looking for experienced, reliable, and qualified Python developers? If yes, you have reached the right place. At **[]( "")**, our full-stack Python development services...

Applied Data Science with Python Certification Training Course -IgmGuru

Master Applied Data Science with Python and get noticed by the top Hiring Companies with IgmGuru's Data Science with Python Certification Program. Enroll Now

Hire Python Developers India

Looking to build robust, scalable, and dynamic responsive websites and applications in Python? At **[]( "")**, we constantly endeavor to give you exactly what you need. If you need to...