Data Science Portfolio: An Employer's Point of View
Data Science Portfolio: An Employer's Point of View
What is the company looking for in your Data Science Portfolio?
In recent years, companies have realised the importance of utilizing their data to gain a competitive edge. Corporations, startups, and small businesses have used data projects to improve operations. This trend has heightened the demand for qualified data personnel, a need that continues to grow.
As this vibrant industry expands and demand increases, many individuals are learning data science to enter the field. This has led to fierce competition. Why is the competition intense despite the growing needs? The area is still relatively new, and many companies prioritize experienced candidates over fresh graduates. This dynamic makes the entry-level data science position a significant bottleneck. So, how can you stand out in this sea of applicants? One strategy is to tailor your portfolio to meet employers' specific needs.
In today's newsletter, I intend to share my insights about what I, as an employer, look for in a prospective applicant's Data Science Portfolio. Let's dive in.
Considering our discussion about Data Science Portfolios, I'd like to recommend my friend Admond Lee Kin Lim's newsletter: Hustle Hub. It's an exceptional newsletter where you can learn tips to advance your data science career, covering everything from data basics, tech tools, recommended books, effective work methods, and more.
If you're aiming to distinguish yourself in your data career, it's a must-subscribe. Check it out below!👇
Structured Portfolio
What is more unpleasant than encountering a disorganized data science portfolio? When the narrative is unclear, and the conclusion is absent, that's often the result of presenting your coding and data exploration projects in their raw, unfiltered state. I frequently observe this issue: people may know what type of data science project they want to tackle and attempt to resolve it; however, they fail to structure their portfolio in a way that allows others to understand the process quickly.
For instance, let's consider an individual who aims to address fraud by developing a model to predict fraudulent activities, ultimately achieving good metrics. This individual managed to land an interview because of their fraud prediction model but failed to articulate a straightforward narrative of how they developed it and how it could benefit the company. Next, the individual didn't receive a job offer because the interviewer felt they could not explain their workflow adequately. The interviewer was concerned that this lack of clarity could translate into a disorganized approach during employment.
This example reflects a common scenario in my experience. People often focus solely on their results without elaborating on the process, lacking structure. Employers are interested in how effective the model you've created is and your thought process leading up to the development of that model. Therefore, creating a structured data science portfolio is crucial to stand out.
The structure can vary, depending on your specific Data Science project. However, it should always tell a complete story, from beginning to end — even within your Notebook. A typical structure includes:
Explaining the problem you want to solve
Describing the data and how you acquired it
Documenting the data exploration process
Detailing the model development process
Presenting results and drawing conclusions
With these five steps, you can create a well-structured data science portfolio. You can develop it using presentations, Git, Notebooks, or any other mediums — as long as it effectively tells your story. You earn extra points if you explain the specific benefits of implementing your project.
Addressing the Business Problem
It's an exciting achievement to develop an impressive neural model that can accurately predict facial features and the clothes someone is wearing. However, does this project solve any real problem that the company you're applying to has? It might, but it's a stretch and likely won't be relevant at the interview stage.
Companies hire data scientists to solve their unique business problems. Job ads often list technical requirements, some domain-specific, but rarely explicitly state the specific issues the company needs to solve. Applying for such roles with a generic portfolio could work, but it's a gamble; numerous applicants might have submitted similarly "typical" resumes and data science portfolios.
So, how do you stand out? Align with what the company needs: solutions to their business problems. Endeavour to create a specific data science portfolio that addresses these issues. For instance, if you're applying for a data scientist position in the financial industry, your portfolio could tackle common problems in this sector, such as investment strategies, fraud detection, risk management, and more.
I appreciate applicants who, at the very least, research the company they're applying to. This effort demonstrates a willingness to invest extra time in understanding the company. It's in the company's best interest to employ someone who already comprehends the business or the company compared to one who lacks this knowledge. Your data science portfolio will stand out if it addresses the company's business problems.
Being Creative
This ties in with how your data science portfolio should address business problems. Avoid using staple projects you learned from courses or books when choosing your project. Instead, your data science portfolio should be built based on your creativity.
Imagine applying for a data scientist position in the banking industry with a Titanic or Iris dataset project. What would that convey?
First, these are projects that everyone has done.
Second, there's no business need for these projects (maybe historians or biologists would find them relevant?).
Third, it could harm your chances as employers perceive you as lazy.
You could undertake numerous other projects to demonstrate creativity when building your data science portfolio. Creativity can be displayed in several ways, including:
The business problem you choose to solve
Your approach to the problem
How do you explore the data and summarize the insights
How do you develop models and troubleshoot technical issues
Crafting a conclusion for your data science portfolio
And many more aspects. Creativity means you aren't confined to what's taught in courses or standard practices but can showcase how you innovate beyond these boundaries.
In my experience, while reviewing applicants' data science portfolios, I examine how they solve the problem they've chosen. Many people fall into similar patterns, but I always get excited when I find someone who can provide a unique perspective.
I recall many applicants using the same dataset in their portfolios and attempting to solve the same problems. However, only one individual stood out. He distinguished himself by summarizing his insights differently. While many people dismissed certain features as unimportant, this candidate conducted a deeper analysis and discovered that the problem was a compounding issue—this insight required creative thinking.
Being creative may require more practice as it's related to critical thinking. However, once you get accustomed to viewing problems from multiple angles, I'm confident that many employers will value your creativity.
Key Takeaway
Data science is a rapidly growing industry characterized by intense competition. To secure a position, you must present a compelling data science portfolio. But what exactly are employers looking for in your portfolio?
In this newsletter, I've shared my perspective as an employer, outlining what I seek in a data science portfolio. The key aspects are:
A Structured Portfolio
Problem-Solving Focused on Business Issues
Demonstrated Creativity
I hope you find this information helpful!
Thank you for reading the Non-Brand Data Newsletter. If you found this helpful post, please share it with your friends. Also, I encourage you to comment on any topics you'd like me to write about!