DARING TO BECOME A DATA SCIENTIST:
A Roadmap on Challenges and Remedies for Transitioning into Data Science
This article was written by Joseph Itopa A. and Dr. Rosana de Oliveira Gomes in collaboration with Omdena.
The advances in Artificial Intelligence techniques and the overwhelming amount of data produced in society daily have opened doors to many opportunities in industry and research. Data Science is no exception, and has become one of the most famous areas when it comes to the tech industry. Regardless if you want to become a data scientist, a machine learning engineer or an artificial intelligence researcher, getting started might seem quite challenging, especially if you are changing careers. This article has to goal to help aspiring data scientists (regardless of changing careers or not) in their path into the Data Science world.
Imagine boarding a plane mid-air. You may start thinking that this might be a suicide mission and that you should jump off and go by train. After reading this article, you will know that changing careers does not need to be that difficult, most especially when Data Science is what you chose to hop into.
Before we start writing about steps you can take in becoming a data scientist, you need to understand that the Data Science profession is relatively new, meaning that most data scientists or ML engineers did not start their careers in this path. They changed careers from other fields! And you can do the same regardless of your field of study. What you will need to do is to make up your mind about how you want your future to be and remain focused.
As every new experience in life, there will be challenges on this new path that you have decided to venture into. The first step should be to draw a strategy to fulfill your career change goals, and one helpful way to do so, is to understand what will probably be expecting you down the road and what measures might be needed. In what follows, we look at your transition into Data Science both from a technical and soft skills points of view, highlighting possible challenges and remedies. Let’s get started.
Possible Technical Challenges and Remedies
1. Brush up your Math and Programming Skills
Every Artificial Intelligence algorithm is written in a programming language and is based on a mathematical structure. This usually involves Linear Algebra and some concepts from Calculus. The interpretation of results is done using Statistical Analysis. In other words: whether you like it or not, whenever you deal with data science, there will be a lot of programming and math involved. This means that you have to choose a programming language to learn, which in this field usually is either Python or R, most of the time combined with SQL and Bash. You can have a look into the results from Kdnuggets poll on programming software used by data scientists on the image below. The choice of programming language is pretty much a matter of style. People already familiar with other programming languages tend to opt for Python, given its similarities. Python has also the advantage of being widely used, making it much easier to find help for solving a complicated problem in online forums or platforms. When it comes to the math required for Data Science, it is important to say that these are the typical initial courses of any science degree, so there is no need to be scared. Nowadays there are several resources for learning these concepts from renowned institutions online (for free!), such as Coursera courses (check these courses 1, 2, 3, 4, 5), MIT lectures or Youtube videos. The golden rule while learning both the math and the programming is: take ONE learning source at a time and stick to it. The worst thing you can do is to keep learning the same things over and over because you are taking too many courses.
The golden rule while learning both the math and the programming is: take ONE learning source at a time and stick to it.
2. Become a Problem Solver
You can see the science of Data Science as the ability to solve problems through creative and logical thinking. This of course requires knowledge of programming and the understanding of algorithms, but in the end, what you need is practice and more practice. After acquiring some basic knowledge of programming, you can start venturing into solving real world problems by practicing on courses or platforms. Some good hands-on sources for getting started with projects are the courses at W3Schools and GeeksforGeeks. At Data36 you can access courses and also purchase a server (U$ 5) from Digital Ocean to practice. Solving some Kaggle competition problems can also give a boost on your problem solving skills, as you can easily lay your hands on data to practice with and find a lot of help in the community. The golden rule here is: look into projects of your interest, both in terms of skills you want to build and subjects that you feel passionate about. It is all about having fun, after all.
The golden rule here is: look into projects of your interest, both in terms of skills you want to build and subjects you feel passionate about.
3. Join Collaborative Projects or Bootcamps
After getting familiar with some algorithms and practice on small projects, you will naturally transition into searching for more advanced projects. This is when Bootcamps and Collaborative platforms come in. Data science bootcamps are usually associated with fast-paced and hands-on programs in which students build skills in a semi-professional setup. On the other hand, collaborative Data Science puts its coins on community collaboration in order to develop projects in a more diverse and productively way. Though the styles of each initiative differ, the goal should be to advance your skills and get familiar with end-to-end Data Science projects. Bootcamps will make you learn a lot from mentors and how to apply your knowledge in a structured project. Their disadvantage is that in most cases the data used comes already clean, making you miss a crucial part of a Data Science project. The ruggedness and fun in street coding is best found in collaborative projects. You will learn from others as they learn from you. You will make friends while struggling with unstructured and messy data. There are several inspiring collaborative initiatives out there, such as Data Kind, Science to Data Science or Data Science for Social Good. However, most of these options are dependent on location, costly or available only once per year (usually subject to extensive application process). One alternative to collaborative projects is Omdena, which launches several projects per year and applies a principle of volunteering to solve real world problems through online collaboration. At Omdena, people work with domain experts thereby putting you at the heart of learning data science. It is quite likely that, as you join a big project, you will become more motivated to keep learning concepts and ideas (from webinars, courses, books, blogs posts, etc) which will help you in the projects. The golden rule: always keep learning, so participate in projects to take a step further and dare to go beyond the things you have already learned. Maybe you know a lot about algorithms, but not much of data cleaning (or vice-versa), so go for it!
The golden rule: always keep learning, so participate in projects to take a step further and dare to go beyond the things you have already learned.
Possible Non-Technical Challenges and Remedies
1. Organization and Time Management Skills
One important thing that you must make clear for yourself is: changing careers is a project. It requires a strategic plan, a timeline and specific (and realistic) milestones to be achieved. This means you will need to start making plans and ask questions about the field of Data Science. Some questions you might need to ask yourself are:
- Why do I want to be a data scientist and what subjects am I passionate about?
- Will I quit my job and take the time to learn the skills I need or will I make the transition in parallel with my current work?
- What am I good at? What are my weaknesses?
- How much time and money am I willing to spend on changing careers?
- What are the new skills and qualifications that I need to acquire for the new career path?
- What is my learning process? Hands-on or analytical?
Although this may sound like basic interview questions they are of great relevance to determine what your strategy is going to be. Once you have established your goals, what and how to learn and how much time and resources to spend, you will still need to organize this in a plan. The more visual the better. If you need to print some sheets with a weekly plan and stick them on your wall, so be it. If you rather have all organized on a daily schedule on a Google calendar, that is also great. In the end, you have to find what works for you. The golden rule: after some time following your data science plan, you will have to come back to your original plan and ask yourself the same questions again. This is how you will identify what is working and what needs adjustments. Maybe you underestimated how much time you needed for a course or maybe the learning resource you chose is not quite as you imagined.
The golden rule: after some time following your data science plan, you will have to come back to your original plan and ask yourself the same questions again.
2. Become a Better Communicator
We started this article talking about how math and programming skills are must skills for whatever path in Data Science. Here is another one: Communication. A data scientist is nothing more than a person who makes sense out of tons of data and brings it out there to decision makers and stakeholder. However, not all the people that you will need to communicate with are Data Scientists or have a background in programming and math. To make it worse, it is quite common (but of course not a rule) that people interested in these skills are introspective and not super excited about public speaking or constant communication. Putting all this together with the fact that you are changing careers and that it is quite possible that you feel unsure of your abilities; you have a quite high chance of experiencing Impostor Syndrome. This is a condition in which highly qualified personshhave thoughts think they are a fraud and are not good enough. Omdena has hosted a webinar on the topic and how to overcome this challenge. The good news is that the one of the tips for overcoming impostor syndrome is also the same as for becoming a better communicator: practice. You are new to an entirely different field and you are ready to learn. This will happen through practice. Communication in Data Science in not only about being a good speaker, but also about providing good explanations of complex concepts to non-experts. It is also about documenting codes, so other team members can learn from your work. It is about being a good writer and getting the habit of writing about everything you do in your career path. Our golden rule here is: documentation is power, so write and keep writing. In the very near future you will need to show all those skills that you put on your resume, and they will be amazingly backed up by having all your work ready to be shared in nicely documented codes (for example, on Github), blog posts and possibly talks about your work. The earlier you start, the faster you will improve.
Our golden rule here is: documentation is power, so write and keep writing.
3. Build Your Network
Last but not least. You will implement your career plan, build technical skills, become a better communicator… and who is going to be there to see all you have done? Though many people feel awkward about the idea of networking on their job hunt quest, there are some important things to be said. First, most of jobs people get come from their network, not from applications (check the references here and here, for example). Second, networking does not need to be something artificial or manipulative as it is sometimes interpreted. Most of the time, you want to connect to people who have the same career you want to have in order to receive advice, ask questions and just learn about how their careers are in practice. People are usually glad to help and excited about sharing advice, and the worst thing that can happen is not getting any response. One of the easiest ways to network is through collaborate projects. There you will share your knowledge with juniors learners, as well as work with more experienced practitioners, from whom you will learn and be in closer contact. These are the people who experience how it is working with you and who will usually be glad to recommend you or keep sharing knowledge with you. Another good way to network is by following Data Science media such as Omdena, Women Who Code, Women in AI, Super Data Science and getting in contact with interesting people you find there. It can be the speaker from a podcast you liked, a teacher from an online course or a blogger whose posts you really enjoy. The last golden rule is: put yourself out there and ask questions. You will definitely get a feedback most of the time.
The last golden rule is: put yourself out there and ask questions.
In conclusion, changing careers is a project in itself and will require hard work. Even if you are still transitioning, you can practice data science on your free time and use the resources available in this article. Being persistent while learning, remaining curious and developing a question-oriented mindset will be an important part of your journey.
While the technical point of view will help you build skills for you to begin in the field, the non-technical resources will offer you the mindset needed for you to take the step of changing careers. Both parts of this development are important, as you are going to use most of these resources throughout the rest of your career. We wish you a happy transition and a successful career!
Both authors are currently transitioning into Data Science from different fields. To contact the authors and continue this conversation, contact us on LinkedIn (Joseph, Rosana, Omdena).
