What exactly is "data science" these days?

Definition

Matt Brems, a leading data science instructor, shares his insights on the evolving definition of data science. He describes it as using data to make informed decisions, highlighting its flexibility across various domains like analytics and business intelligence 1. Matt emphasizes that data science is not just about technical skills but also involves subject matter expertise, as illustrated by Drew Conway's Venn diagram 2. He notes, "Data science is represented as the intersection of math stats, computer programming, and subject matter expertise" 2.

   

AI & Data

The intersection of AI and data science is a focal point, with Matt stressing the importance of foundational data skills before diving into advanced AI techniques like neural networks 3. He points out that without quality data, even the best AI models will fail to deliver desired outcomes. Chris Benson adds that many newcomers are drawn to the allure of AI without realizing the significant groundwork involved in data preparation 3.

If your data isn't good, it doesn't matter if you selected the world's greatest neural network, it's not going to do what you want it to do.

--- Matt Brems

This underscores the critical role of data science in the AI workflow.

   

Toolkit

The data science toolkit has evolved significantly, reflecting changes in technology and educational approaches. Matt discusses how tools like TensorFlow and GPUs have become integral to data science education, alongside traditional tools like pandas and scikit-learn 4. He explains the challenge of standardizing curricula for students with diverse backgrounds, emphasizing the need for pre-work to level the playing field 5. "Our pre-work is an attempt to get folks who may not be at that level yet to prepare before the program," he notes 5.

   

Specializations

Emerging specializations in data science are shaping hiring trends, with SQL skills being highly sought after 6. Matt observes that while SQL is essential, it is often considered a basic requirement rather than a differentiator. He suggests that companies might benefit from upskilling existing employees, as it is often more economical and less risky than hiring externally 7.

It's generally much more expensive to hire someone new as well as riskier than to train someone up internally.

--- Matt Brems

This approach leverages internal knowledge and aligns with the evolving demands of the industry.