#MakeoverMonday Week14: A Journal on How I Created my Submission for the Week

The task was to create a visualization to better represent the characteristics of the different classes of beans. If you want to play around with it, you can find the dataset here.

This data is made up of mostly numerical columns. So at first, this threw me off balance but after spending time on it, I thought a bump chart would be a great way to represent all the tons of numbers in the dataset. The initial visualization is below:

What Works Well

• Seeing the different varieties of beans in a glance. The classification rate for each machine learning method used is evident from this visualization.

What Could be improved

• Color reduction: all the colors were just in my face and they were a bit too much. In their defense, though it was relevant to pass the message.
• The gridlines are chart junk and can be done away with.
• More description of the visualization to explain what the visualization is about. Like a title will do everyone a lot of good.
• I have reservations about the axis starting from 80, it really does not distort the message being passed but I still have reservations about it.

My Visualisation

Thought Process: I wanted to see everything at a glance. Sometimes, this is not always a good idea but I always think that it won’t hurt anyone if my audience can see the story I am trying to tell without having to click too much.

I decided that it would be a good idea to make the data tall as opposed to wide, hence my decision to pivot the table.

The topic is one most people won’t be familiar with, therefore it was essential to making the heading captivating while also giving supporting text to the subject in question.

The bean seed when hovered on gives you a definition of each characteristics in the chart. Click this video to find out how I achieved that. It was actually pretty simple. The only challenge I had here was trying to make all the bean seeds the same size since the size was irrelevant to the visualization.

Check out the Interactive Viz on my Profile.

How My Dashboard Was Achieved

1. Pivoted the table and placed ‘class’ and ‘value’ columns from the dataset in the row and column section on my worksheet. Also placed the measure column from the data in the column section of my worksheet.
2. Moved class from the column section to the details in the marks card.
3. Edited default property of measure to AVG
4. Selected rank from quick table calculation for the Value pill as opposed to using the sum of the values.
5. Changed marks type to circle as opposed to using the automatic mark.
6. Edited table calculation to enable rank to be calculated using the measure field in the data.
7. Duplicated the value pill to create two different sets of columns
8. Edited the table calculation and set to calculate rank by class this time.
9. Select the dual-axis for the duplicate of the average pill.
10. Swapped rows and columns.
11. Reversed the rank order to reorder the rank.
12. the rest of the analysis was really basic formatting and removing any chart junk. At this point, you can do whatever feels most appealing to you.

P.S: I give full credit to one of the most amazing Tableau Genius I know @Andy Kriebel. Check out the video here. This chart can be found from 37:29 — 41:04.