Data Vizualization/Dashboard
Project 1 - Interaction Design ll course at SJSU
Challenge
Translate large data sets into a design that helps fraud detection managers monitor, track, detect and solve issues related to performance of employees, queues and models.
-
Issues should be immediately detected and understood
-
Provide sufficient information to help make decisions and come up with effective solutions
Goal 1
Information architecture should support:
-
Search
-
Filtering the data
-
Drilling down on the data
-
Tracking data
-
High level aggregate information
Goal 2
Dashboard questions to be answered:
-
Analyst performance
-
Queue status & performance
-
Model performance
Design Process
Analyze
Understanding the user experience in its entirety through journey mapping
Empathize
Truly considering user needs, wants and goals through user stories and personas
Define
Developing a conceptual model that communicates what users can do and how
Ideate
Sketching and creating flows to come up with the best solutions before prototyping
Prototype
Translating all the information from previous steps into an effective user interface
Analyze
Journey map
For my first solo project, I generated the entire user journey for the main persona on my own
I focused on thoughts, feelings and actions throughout the process to understand critical wants and needs in relevant context
(pg 1 of 3)
Takeaways:
-
there are many competing tasks and required information that can be difficult to sift through along with the time pressure of fraud detection
-
people can be difficult to manage, there are many variables to consider and how those variables relate to Janine's performance as a manager
Design Goals:
-
prioritize and emphasize pertinent information to reduce cognitive load and processing times
-
give Janine the information needed to make decisions, don't just tell her what is wrong and where, provide context within the bigger picture
Empathize
Persona
Typical Workday
Run scenarios that allow quick response to changing customer/applicant behavior, as well as response to changes in results as new data becomes available.
Janine
A Perfect Day
Instead of having to react to operational challenges in a knee- jerk fashion, she is able to proactively run scenarios that consider likely challenges, so that she has a library of scenarios and solutions ready to quickly deal with operational challenges
Pain Points
• Difficult to get reports to see the “big picture”
• Hard to find the data she needs
• Slowed down waiting for time from IT to add data sources
• Reports don’t provide visuals to communicate clearly to upper management
• System performance, number of different/disparate systems needed to make a decision
• Productivity loss due to inadequate tools and training
User Stories
With the user journey and persona established, I sought to express user goals through stories that include objects, actions, and attributes which later become inputs to the conceptual model.
Takeaways:
-
Past, present and future information is important in specific contexts. Consider these contexts to leverage the hierarchy of information
-
Janine wants to immediately and easily identify the issues upon login and dig deeper when needed to uncover possible solutions and act on them all at once
Design Goals:
-
Give Janine yesterday's stats in relation to what is currently going on and highlight possible similarities and trends between the two
-
Relevant information is readily available during tasks to support accurate decision-making
-
Help Janine find relevant info needed for the task, and reduce the cognitive load required to complete tasks as much as possible.
Define
Conceptually, every product can be simplified down to a set of objects, attributes and actions. For example, fraud cases are objects, case statuses (open/new/active) are attributes, and monitoring those cases/statuses are actions. These objects, actions and attributes directly translate to the Nouns, Verbs and Adjectives from user stories.
Object/Action
Contrary to previous projects, this design included fewer objects and actions.
This allowed for more detailed focus on fewer items, which seemed like an simpler task at first...
Object/Attribute
...I soon became aware of the complexities involved with data viz.
The number of attributes per objects exploded the bigger picture onto a much broader canvas of complexities and requirements.
Prioritization
My design method helps synthesize and consolidate the mass information from the attribute table to a more precise hierarchy.
Below is a guide to build the information architecture based on user research established in previous steps.
Takeaways:
-
Fewer objects with more attributes meant less screens with more details in each screen
-
Important information can easily become lost in a sea of details
Design Goals:
-
Frequent actions need to stand out among the potential noise
-
Provide Janine with all the information needed to make accurate decisions, but make monitoring and managing pop
Ideate
Sketches
My main goal was to first generate a high level layout of where to place information on the screen that makes sense at first glance. This required strong understanding of user goals and geometrical expression of my conceptual model. Drawing up the charts allowed me to quickly judge the effectiveness of what I was trying to communicate in each screen/section- a valuable time saver.
Takeaway:
-
there are NUMEROUS ways to visualize data
-
the expression that best aids user goals in this context included ones that communicate speed and accuracy of analysts and queues
Design Goal:
-
use line charts to communicate trends over time - allow control of time period
-
incorporate brushing (highlighting desired information across entire screen) to quickly reveal patterns and potential issues
-
highlighting potential issues is half the battle, include info on why those issues arose and how to effectively solve them
-
allow high and low level info processing through zoom, filter and detail tables to show users the whole picture down to the pixel
Prototype
Below is a collection of my most innovative and impactful solutions to the design problem based on information gathered in previous steps. Check out the prototype before and during the explanations here
Home
Everything fits without scrolling to enable "big picture" processing
Yesterday's tallies are presented first with the ability to push to bottom
Information organized by objects to establish consistency between user mental model and conceptual model
High risk items are highlighted in red to identify patterns and address immediate issues
Information can be pre-attentively processed vertically and horizontally
Summary
What I learned:
-
Keep in mind the persona - just because I came up with a quick and clever visualization doesn't always mean it helps the user solve common goals
-
Think outside the box - there's a million ways to visualize info, don't stop until you've exhausted every idea (but keep it low fidelity until then)
-
Visualizing information is not just making the right content pop (although important), its about letting users dig deeper and supporting decisions