The world’s leading publication for data science, AI, and ML professionals.

3 Things Not to Do When Learning Data Science

Avoid them to reach your goal faster.

Photo by NeONBRAND on Unsplash
Photo by NeONBRAND on Unsplash

In recent years, companies realize the potential of Data Science more than ever. They do major investments for data-related solutions. As a result, the data science ecosystem has experienced tremendous growth.

It is inevitable that such a popular field attracts people as a career option. Many people from different professions make a career change to work in the field of data science.

However, this is not a smooth and easy transition. Data science is a highly broad field. It requires lots of learning and adapting many software tools. Thus, you need to work hard and dedicate yourself to successfully complete this transition.

I am one of the many who decide to make a career change to become a data scientist. It took me almost two years to land my first job. I have written many articles to share some details about my two-year-journey.

In this article, I will write about 3 things that I think aspiring data scientists should avoid. They have the potential to break your motivation or slow you down.


Do not try to compete on Kaggle

Kaggle is a great platform for learning. You can find extremely valuable content there. The notebooks shared by people contain great learning material, the datasets can be used for practicing, and so on.

I think, however, the competitions are not for people who are new in the field. I’m talking about the competitions with a prize. There are some playground ones which are great for practicing and learning as well.

The ones with a prize are very difficult. You can still learn a lot of things while competing on Kaggle. However, as a beginner, it might break your motivation for learning.

The teams are likely to spend long hours in order to gain a tiny bit of accuracy improvement. At the beginning of your data science career, you should be spending that time on learning more fundamental concepts.

Even if you spend a ton of time on those competitions, you might not even get close to the leaderboard. This is absolutely normal. However, not getting better or at least average results might cause you to feel inadequate. You do not want that at the beginning of your career.

Consider Kaggle as a great platform for learning, not a place to test your knowledge or skills.


Do not fall for Python-R dilemma

There is a ton of software tools and packages in the data science ecosystem. These tools help us complete the tasks seamlessly and efficiently in most cases.

The benefits of these tools are not open for discussion. However, they might turn into a disadvantage if not used wisely.

In many cases, you will have multiple tools to perform a task. The rich selection of tools often lead to discussions that involve some kind of comparison. For instance, you must have read at least one article about Python and R comparison.

Just pick one and go! At the beginning of your career, it would be a waste of time to question which one to choose. Most of the options will be enough to learn the basics.

This is not limited to the Python-R comparison. For instance, Matplotlib, Seaborn, and Altair are three different data visualization libraries for Python. I’m pretty sure anyone of them will satisfy your needs.

You can do data analysis and manipulation with both R and Python libraries. Which one you choose does not matter, at least when you are learning. Once you get your first job, you can make better decisions on your preferences.


Neural networks sound cool but…

Deep learning is a subfield of Machine Learning. It is used for solving machine learning problems with neural networks. The principals do not change. You create a model, train it, and evaluate it based on a loss function. This is an iterative process so you tune your model to improve its performance.

Deep learning algorithms are more complex than traditional machine learning algorithms. They are likely to perform much better in some specific tasks. For instance, a convolutional neural network is probably the best choice for an image classification task.

However, many problems that are in the broad range of data science do not require using deep learning models. The machine learning algorithms are more than enough in most cases.

Neural networks sound cool but do not spend too much time trying to learn them at the beginning of your learning journey. You can always learn them if and when you need.

Data science is such a broad field with lots of different applications. Depending on your job, you may never need to use neural networks. Besides, if a problem is solved with a simpler model, no one will force you to use a deep learning model.

You should focus more on learning the statistical concepts related to the machine learning algorithms. Improving your statistical knowledge will definitely help you a lot.


Conclusion

It takes hard work and dedication to become a data scientist. The hardest part is to land your first job. Once you are in the field, the rest is relatively smoother.

The learning journey until you find your first job is already a tough process. On top of that, the three things I mentioned in this article might further slow you down. I think avoiding them will be of your best interest.


Last but not least, if you are not a Medium member yet and plan to become one, I kindly ask you to do so using the following link. I will receive a portion from your membership fee with no additional cost to you.

Join Medium with my referral link – Soner Yıldırım


Thank you for reading. Please let me know if you have any feedback.


Related Articles