BENCH BLOG 1 - Visual Analytics

by Mohammed Uddin

   When thinking about visual analytics; the number one thing I keep in mind is understanding. At a glance, I want my audience to quickly grasp what my work is saying. Yes further questions will be asked, but there’s less confusion in the air. So how do we go about this? Well the job is less daunting than it seems!

Visual Best Practices

One method when formatting is to start from the biggest and end at the smallest. For example, if we format a workbook first, there are less changes we have to make in the future. Next, let's focus on color. Changing color in Tableau is a matter of clicking on the color of the marks card. My mindset when using color is to use more muted hues. This means to make sure the colors are not too distracting; taking the audience away from the analysis. Another rule I follow is to make all colors uniform. If my profits are blue, my sales will be blue. This implies anything relating to revenue are related to each other.


Another method is changing the font; it’s best to choose something that’s easily readable.  E.G.:

  • Arial
  • Trebuchet MS
  • Verdana
  • Times New Roman
  • Lucida sans

Another method is changing the tooltip.

To highlight what’s important.       Or to tell a mini-story.             Or to rewrite as a sentence.

When it comes to changing axes, it really depends on the chart that you are making. Synchronizing your axis when using a dual axis makes for an easier interpretation of the data.

Visual Analytics Best Practices - Aiming to be a better storyteller.

Here’s a good example of visual analysis. Although like anything with life, things change. Areas to focus on are:

Representing Data for Humans

When we look at things, our visual mind emphasizes one thing over another. As seen in this picture below.


We like time to be on an axis, location on a map, comparing values using a bar chart, exploring relationships on a scatter plot, and having relative proportions on a treemap.

We want our final product; the dashboard to pass the five- second test. Here the most important view goes on the top or top-left, the legends go near their views, best to multiple color schemes, use 5 views or fewer in dashboards, and provide interactivity.

Which Type of Chart or Graph is Right for You?

When talking about charts in Tableau we essentially have two types.

  1. Bar Charts

**Horizontal** is preferred

- Because it’s reads like a list.

- labels are easy to read

**Vertical bars**

- Reserve for time

- When you are comparing time.

- Each bar is an individual component

- Never NOT START AT ZERO

- SO START AT 0

**Histogram**

- Numerical categories/buckets

- BINS are evenly-sized (Right-click dimension and change bar chart)

**Grouped/Paired**

multiple  comparisons in te same chart

**Bar in Bar**

- Direct comparison with the aim of progress

- This year vs. Last year and a line determining good vs. bad

**Stacked Bar Chart**

- Shows segments that make uo the bar

- Bar is sliced up by color

**100% Stacked bar chart**

- When you want to see all the components of the car

- Pie Alternative

**Dot Plot**

- Direct alternative to the BAR chart

- Use when too many BARS

**Jitter plot** adds space  =

- showing the detail

- if confusing we  turn into a histogram

- Bee swarm

  1. Line Charts

- rending time

- Better for uneven/gaps in time

- Size of your sheet will change how your chart is interpreted

- Truncated axis = doesn't start at 0

**Stepped Line** - useful when there is an ABRUPT change

- not easy for people to understand

- recommend to just make a bar chart

**Dot marks** - included when there is missing time

**Curvy Lines** - less precise reading/interpretation

**Stacked line chart**

- Advised not to be used, because its hard to explain

**Slope Plot/Chart**

- Before/After

- The in-between doesn't matter

**Bump Chart**

- Change in RANK over time

- magnitude doesn't matter

**Cycle Plot**

- Mechanical, manufacturing, and weather

Why the Time-Tested Science of Data Visualization Is So Powerful

   Data Visualization has been implemented by humans for generations. Whether it be the inhabitants of the Marshall islands and their boat navigation or Florence Nightingale explaining to “government officials how treatable diseases were killing more soldiers in the Crimean War than battle wounds”. It’s been part of the human experience and it will continue to be. Although the methods have changed from carvings on a wall to a 13 inch Dell laptop; the intuition has not. We all want people to understand us better; visuals are one of the best methods to do that. Whether it be for Iron Viz or a bank client; you're not just a cog because your work matters.