Best Way to Aquire Data Science Skills
Think a career in data science is right for you? Here are four steps to acquiring the skills you鈥檒l need to land your first job in this high-demand field.
1. Love the Data
It鈥檚 the prerequisite to diving deeper into the field of data science. Unless you have this step nailed, steps #2 - #4 can be a long, joyless slog. But if you have a passion for figuring out how to solve problems by looking at the numbers, all the hard work and learning in your degree program will be a delight along the way.
When you鈥檙e halfway through your data science program, staring blankly at a page full of formulas and confused about neural networks, it鈥檚 the love of finding and visualizing answers that will see you through.
So how to get started?
You鈥檙e first step shouldn鈥檛 be learning advanced mathematics or programs like Python. Rather, find something you鈥檙e interested in, something that you鈥檙e curious about, and start asking questions like 鈥渨hat kinds of patterns can I see?鈥� or 鈥渨hy does this trend occur?鈥� Then, start looking for available data to answer these questions. You don鈥檛 have to do any fancy analysis yourself鈥攋ust start by looking at other people鈥檚 analysis and getting jazzed by what is possible.
Here are some ideas to get you started.
- Mapping U.S. flights the day before Thanksgiving
Check it out, play around with the numbers, and see what gets you excited.
2. Learn by Doing
Learning the theory of data science is essential鈥攜ou have to know the underlying statistical principles of your models. But it鈥檚 in the hands-on doing that you acquire the most skill and gain an understanding of what a career in this field will be like. It鈥檒l also give you a nice portfolio that you can share when you start interviewing for a position.
In Loyola鈥檚 data science master鈥檚 program, our course outline emphasizes hands-on application. For example, in the course Machine Learning (CS737), students complete a final project of their choice. In recent semesters, we鈥檝e had one student work to predict the type of beer from available information about beers, and another try to predict whether Kickstarter projects would be successful. Students identified projects based on their hobbies, work, or something they found that is interesting on Kaggle, a Google-owned platform for predictive modeling and analytics competitions.
Outside of the classroom, don鈥檛 be afraid to work on your own projects, too. Simply finding a dataset that interests you, and then using tools like Python to answer interesting questions about it, will teach you a lot about data cleaning, modeling, and visualization.
3. Share Your Insights
Once you launch your career, you鈥檒l constantly have to share your findings with others鈥攑eople who aren鈥檛 data scientists, aren鈥檛 statisticians, and don鈥檛 really understand the data magic that you do.
Communicating your insights clearly, in a way that鈥檚 not condescending and is actually interesting and engaging, can set you apart from other data scientists, get you noticed, and boost your career. This means understanding the topic inside and out, figuring out how to clearly organize and visualize the results, and being able to explain what the results mean and why anybody should care.
Might as well start practicing now. Here are some ideas.
- Share project insights with family and friends. Can you hold their attention for 5-10 minutes and help them understand what you鈥檝e discovered?
- Post results on a blog. Share on social media, especially among groups who might be interested in the topic of your project. See who 鈥済ets it鈥� and also learn from the questions that people may ask.
- Join communities like and the .
4. Level Up
Either in your degree program or in the workforce, if you get to the point where you鈥檙e completely comfortable with the project you鈥檙e working on and aren鈥檛 learning anything new, it鈥檚 time to level up. If you aren鈥檛 learning and growing in this field, you鈥檒l be falling behind.
Here are some things you can do to challenge yourself, both now and throughout your career:
- Try to make your algorithm faster
- Dive deeper into understanding the theory behind your algorithm
- Scale your algorithm to multiple processors
- Work with a larger dataset
- Teach someone else to do what you鈥檙e doing now, as well as the theory behind why you do things that way.