In summer 2017, I interned with a data science team at a consulting firm, focusing my efforts on a machine learning project with two other interns. Looking for ways to combine my data and design skills, I delved into the world of data visualization to create internal dashboards and external reports using libraries such as D3.js, Dash, and Bokeh for the first time.
The main idea behind our summer project was to create a scalable data analytics pipeline that took advantage of both the availability of consumer-grade wearable technology and the power of machine learning. As a part of this project, we also wanted to create a live demo that could be on display at the IoT lab at our firm, and thus we needed some sort of visual that could explain the work we did that summer to place alongside the demo.
Inspired by this fantastic visualization project that gives a basic introduction to machine learning, I wanted to create a scroll-based data visualization that summed up our summer. I chose "scrollytelling" as the way to go for a variety of reasons.
First off, this format really helps one follow along the story of the data, and what we did with it that summer. Because sections flow seamlessly into each other, a reader can easily follow the journey the data takes through our analytics pipeline, a more technical, jargon-y concept that might be a bit harder to grasp for the layperson without proper presentation.
Secondly, a scroller, as opposed to a stepper cuts back on the amount of decision-making the reader has to make, and instead allows the reader to continuously follow the report until the very end, without interruption. This type of format definitely stands out against the other live demos in the IoT lab, which utilized more conventional presentation formats.
I really enjoyed having the opportunity to combine both of my interests in data and design and was amazed at how harmonious these two areas are together. These types of projects resonate with me the most, as the skills needed for them completely match mine. Having knowledge of front-end tools and languages (CSS, HTML, D3.js), an understanding of data science methods and machine learning techniques, and strong design sensibilities all helped me create this visualization!
Take a look at the visualization in full here.