Day 4 of 180 Days of Data Viz Learning

I’m learning data visualization for 180 days straight because I need to #JFDI.  See post explaining how and why I’m doing this.

Today was bit of a break day for me since I powered through the last three days.  I learned in grad school if I’ve done straight learning tasks for four-five hours a day, I start to burn out.  I’m moving to NY for my job next week, so I spent early part of the day working on the inevitable packing before settling into code learning and creation, spending less time on the reading and visualization decomposition today.  I only did one instead of the three I try to do daily, although I only post one on the blog because I handwrite the rest because it helps me learn.

Waking up exhausted today both from the long study sessions from the previous days and an activity-filled weekend (meeting up with friends and upping my workout), reminds me of how I’ll need to manage my study schedules and expectations when I start my job in exactly two weeks.  The great thing is I’ll be doing this all day at work, so I’ll figure out how to allocate tasks to before and after work.  It’s likely I’ll scale back on the coding and creation, and work on the decompositions and outside reading more at that point, but we’ll see!

Tomorrow I’ll need to allocate more time/energy to code and creation work early on during the day before I pack.  I’ll also make tomorrow a rest day in terms of working out too.  In my time at grad school, I’ve come to realize that have the mental energy to focus deeply on pure learning rather than say the ebbs and flows of a normal workday become crucial.

Visualization Worked On or Created: 

My number one priority for today was a bust.  This is the kind of task I want to have hours with to myself, which I’ll have tomorrow.  I did succeed in making some dummy charts in D3.js, which is a small victory.

Decomposition of a Visualization:

Greece’s Debt Crisis Explained

  • What are the:
    • Variables (Data points, where they are, and how they’re represented):
      • Gross government debt as % of GDP over 2014Q4 plotted on single lines of EU countries comparatively, Greece’s GDP ’00 to ’14 Histogram, 2015Q1 Unemployment Rates in Europe % bar chart , Breakdown of Greek Government debt in donut chart
    • Data Types (Quantitative, Qualitative, Categorical, Continuous, etc.):
      • Quantitative Numerical data
    • Encodings (Shape, Color, Position): Position, Color
  • What works well here?
    • Simple no frills chart comparisons.  Good integration of multiple types of narratives (text, data viz, video news story).
  • What does not work well and what would I improve?
    • Donut chart is basically a pie chart that might be better represented with bars given the number of data points represented for scale, but doesn’t work badly at all here.
  • What is the data source?  Do I see any problems with how it’s cited/used?
    • Eurostat, Deustche Bank, IMF, Reuters, Bloomberg
  • Any other comments about what I learned?
    • I think adding one more visual about what the debt could look like when repayments are set to begin would tie the narrative more neatly together.

Code Learning:

Udacity 1b D3 Building Blocks – section in progress

  • Wow Javascript syntax is a pain.
  • Method chaining allows you to easily apply multiple operations to same elements.
  • We can set the svg element’s size in JavaScript so that we can compute the height based on the size of the dataset (data.length). This way, the size is based on the height of each bar rather than the overall height of the chart, and we ensure adequate room for labels.

Reading and Learning Data Visualization Theoretically/Critically:

Visual Information Seeking: Tight Coupling of Dynamic Query Filters with Starfield Displays. Christopher Ahlberg and Ben Shneiderman (PDF) from 7 Classic Foundational Vis Papers

Three Takeaways:

UX design principles that consistently lead to high levels of satisfaction include:

  1. dynamic query filters: query parameters are rapidly adjusted with sliders, buttons, etc.  The effects combined with simple AND OR logics does the job.
  2. starfield display: result sets are continuously available and support viewing of hundreds or thousands of items.  Meaningful two-dimensional displays can be created by selecting ordinal attributes of items and using them as axes.
  3. tight coupling: query components are interrelated in ways that preserve display invariants and support progressive refinement. Specifically, outputs of queries can be easily used as input to produce other queries (output-is-input as a way of efficient usage).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s