5 Tips for Building a Data Science Portfolio

Image by Author

Having a project portfolio today is just one of the standard requirements for data scientists. While we can huff and puff because of sometimes ludicrous things employers want from data scientists, we can’t do anything but try and do better what is required of us.

Data science portfolios are here to stay and without a reason. Actually, I can think of at least five reasons why portfolios are not that crazy of a requirement.

To fully use the potential of a data science portfolio, you should carefully choose projects following these five tips.

1. Do Your Passion Projects

 
Whatever type of projects you do, make sure they are about something that excites you.

Why It’s Beneficial: No employer can stay cold on seeing you demonstrate a genuine interest in data science and the subject. Showing your (professional) authentic self will help you stand out, introduce yourself, and show your creativity in applying data science techniques.

How to Achieve It: Find project suggestions that align with your interests, find publicly available datasets, or collect your own.

Example: If you’re interested in protecting the environment or are a sports buff, analyze environmental data or sports statistics.

2. Demonstrate a Range of Skills

 
Your portfolio should be well-rounded, meaning that it should demonstrate all the main skills required from data scientists.

Why It’s Beneficial: Employers want candidates who are versatile and proficient in all stages of an end-to-end data science project.

How to Achieve It: Think strategically about the projects you choose and the skills they focus on so that the portfolio as a whole demonstrates the required skills. As a segue to Point 3, do several ‘big’ projects that cover all the skills rather than plenty of ‘small’ projects, with each focusing on a particular skill.

Example: Select projects demonstrating your ability to clean and preprocess data, analyze it, build and deploy machine learning models, and visualize insights. For example, try this Prediction of Stock Price Direction.

3. Focus on Quality, Not Quantity

 
Instead of overwhelming employers with the project abundance, impress them with a few high-quality ones.

Why It’s Beneficial: Having fewer well-chosen (see Point 2) projects that are well-executed, polished, and complete makes a good impression. This will demonstrate your thoughtful approach and attention to detail while having a large number of projects done half-assed will only show your superficiality.

How to Achieve It: No matter what projects you do, make sure they are well-documented and you explain your process, methodology, and results thoroughly.

Example: Instead of doing several small projects like simple data analysis or basic ML, you could do a single project that could include detailed data cleaning, exploratory data analysis, clustering techniques such as K-means or DBSCAN, and detailed insights. A good example is this Credit Card Clustering project or this end-to-end project.

4. Highlight Practical Applications

 
Select projects that have practical applications or tackle industry-relevant challenges.

Why It’s Beneficial: Employers like to see how you solve real-world problems because, guess what, you’ll be solving them at your job. In doing so, you’ll be contributing to the business’s goals.

How to Achieve It: First, make sure that your projects address real-world problems. Then, highlight your project’s practical aspects and impacts, such as how using your solution could save time, reduce costs, improve decision-making, and so on.

Example: The Predict Crashes and Facade Risk projects give you a good example of how to highlight practical applications.

5. Contribute to Open-Source Projects

 
Include in the portfolio your contributions to open-source projects.

Why It’s Beneficial: It gives you the opportunity to contribute to real-world applications and demonstrate your ability to collaborate with people from your field. It also exposes you to different coding standards and problem-solving approaches while giving you the opportunity to build a reputation with the data science community.

How to Achieve It: Contribute to open-source projects on GitHub.

Example: Fix bugs in pandas, improve its documentation, or add new features to it.

Source Suggestion

 
Here are several suggestions for where to start searching for publicly available data, data projects, and open-source data science tools.

Publicly Available Data

Data Science Projects

Open-Source Data Science Tools

Conclusion

 
In today’s job market, overcrowded with data science applicants, it’s difficult to stand out. While there are probably numerous excentric ways to do so, I’m for a ‘conservative’ approach here: stand out with your high level of skills, thoroughness, and overall quality of work.

If this is your way, use the tips above to build your data science portfolio and impress employers.
 
 

Nate Rosidi is a data scientist and in product strategy. He’s also an adjunct professor teaching analytics, and is the founder of StrataScratch, a platform helping data scientists prepare for their interviews with real interview questions from top companies. Nate writes on the latest trends in the career market, gives interview advice, shares data science projects, and covers everything SQL.