Three Ways to Use Imagination in Data Science

“Imagination is more important than knowledge. Knowledge is limited. Imagination encircles the world”              

– Albert Einstein

Everyday my inbox fills up with new articles and posts on R tutorials, data viz tips, and the latest and greatest tricks to increasing my modeling predication accuracy. I enjoy reading many of these articles, but often I find that only focusing on the “hard” skills leaves me in a rut.

As Albert Einstein said, knowledge is limited, therefore if we only focus on building our knowledge base we limit ourselves as data scientists. In order to go beyond our limited knowledge base, we must integrate our imagination into our work. Below are three areas I’ve found applicable for using imagination in data science.

Hypothesis Development:

The mere definition of a hypothesis is a supposition or proposed explanation, in order to create a proposition one must think of an explanation not yet materialized. Therefore, from the very beginning a data scientist must use imagination to come up with a theory or hypothesis.

Feature Selection:

Sifting through large amounts of data to find the signal in the noise can be time consuming and laborious. However, if you take the time to imagine what fields might be useful in your predictions and think outside of the box to find these not so obvious indicators, you not only save yourself time but also make better predictions.


Imagination can also be used to see beyond the initial application of your analysis. By using ones imagination to think beyond the initial findings, one increases the application of their work and makes connections for future analysis.

So now it is time to warm up your imagination by listening to your favorite song, drawing a sketch on a sticky note, or busting out a dance move. The more we begin to practice incorporating creative imagination into our daily practice, the more we will be able to see our work go beyond mediocre discoveries and into something extraordinary.