Navigating the data lake

Smart data, big data, red data, blue data. Data these days comes in countless forms and types. Its storage forever-home: the data lake. How might a software program help analysts navigate and process the raw information that sits in these lakes?


For SAS Institute
, Aug. – Sept. 2015

TEAM

Bethany Faulkner
Melanie Loff-Bird

SHARED ROLES

Interaction design
User experience design
Keynote animation

OBJECTIVE

Working with SAS Institute, we asked: How might we design an intelligent and self-learning data management system?

Change the equation

The current data analysis process isn’t pretty. More time is spent prepping the data for analysis than actually analyzing it. Reversing this equation, by inventing smarter, more intuitive methods for a software program to handle data, was our mission.

Data equation diagram
calypso project brief and goals

RESEARCH

Identifying an opportunity area

After mapping SAS’ data management project cycle we identified three main user groups that overlapped throughout the project cycle. Based on the variety of tasks a marketing analyst, one of three user groups we identified, was asked to complete we decided this user’s tasks offered the greatest opportunity for intervention.

Precedent studies

Studying similar products gave us insight into both preexisting solutions and potential places for intervention. Some competitor programs excelled at providing visualizations at all scales, while others too often relied on spreadsheets to execute commands.

user journey map

There and back again, a user’s journey

To understand the data analysis process within SAS’ current software, we constructed a user journey map, illuminated some clear pain points occurring in the middle of the process. Tedious cleaning, partitioning, and transforming tasks dragged a user’s enthusiasm down, and their mood was slow to recover.

task flow diagramtask flow diagramtask flow diagram

From experience maps to explicit tasks

We translated our user journey map into a series of three, explicit task flows. These task flows combined our knowledge of the current process with ideas for intervention.

lo-fi wireframes of each task flow
sketches

INSIGHT

Although marketing analysts clean data on a regular basis, and are the least-trained at this task compared to their peers, they have to work within a program built for experienced data scientists.

Build a data cleaning software for the rookies

We decided to focus on newly-minted marketing analysts, and build a data cleaning software suited to their new needs. This meant it had to be intuitive enough to cut down our audience’s data cleaning time, but without sacrificing the accuracy of higher-level software.

ui iterations

Our final prototype

In this task flow Juno Allcott, a new marketing analyst at a small pharmaceutical company, finds and analyzes relevant data sets about U.S. Lyme Disease patients; she must explore and clean this data in conjunction with her team.