MM CW6 – Digging deeper into single data points of a data set

The target of this week’s makeover is the following viz:

What I dislike about the viz:

  • The choice for a stacked bar chart. The stacked bar charts create a white space between the years. This elicits the impression that there might be missing data in between the years which is not true. An area chart would have been the default design choice to show the relative change of the percentages over time. This would have fixed the white space issue.
  • The color legend is center of the viz and blocks the view. No information is lost by locating the color legend on the red bars for the white ethnicity, but the design feels off.
  • The color choices are counterintuitive. The black and white ethnicities are represented by the colors green and red. This could lead to increased cognitive load.

My Makeover:

MLBDash

The key message I want to convey

  • Only as late as in the mid-1960s, did the first Asian baseball player debut in the MLB. I had to think for some time, what to do with this viz. The intuitive thing would have been to just convert the stacked bar chart into an area chart and show the increased diversity in the MLB or highlight the increase of Latino players. However, I became curious about Asian proportion of player since the line for the Asian ethnicity starts long after the other 3 ethnicities. I did some Wikipedia research and discovered some fascinating facts about the first Asian player in the MLB who is responsible for the first bump in the prior flat lane of Asian players. Originally, I also wanted to annotate and enhance the backstory of the second bump in 1994. But unfortunately, I was not able to find additional information about the player who caused this bump.

The design choices I made:

  • I opted for a simple line graph and filtered out the other ethnicities to magnify the small increase in the Asian Player pool. 
  • I chose the blue of the MLB Logo as a background color. For the color choice, I am still not 100% sure if this is the way to go since I do not really like this color scheme. But I did want to create some resemblance to the MLB Logo, so I stuck with it.

First steps in the DAND and prep

I purchased the DAND (Data Analyst Nano Degree) today. The first section will cover data analysis with SQL and Python. There is an official start date for the cohort which is in 12 days on the 13th of February. Till then I will refresh my statists knowledge with the free intro to descriptive statistics course. As I understand it, the course is itself part of the Nanodegree syllabus so I might be able to already make up some time and work ahead.

Additionally, I started the R Programm A-Z course on Udemy. Kirill Eremenko teaches the course. I purchased his Advanced Tableau course on Udemy as well and found him to be a fun and enthusiastic teacher. I have no prior experiences with R so I hope completing this course will equip me with basic knowledge of this tool and enable me to leverage its advantages which makes it so popular in the data science community.

I will also try to solve one python quiz on codewars.com to expand my python coding skills.

MM CW5 – Time for my first radial bar chart

The target of this week’s makeover is the following viz:

img_2443

First thought, wow there is a lot going on. Where should I bring my attention to?

Which brings me to my next rubric:

What I dislike about the viz:

  • There is too much clutter. There are 5 layers of information. Company logo, company name, fortune rank, net income and profit per second. Company logo and company name are redundant – they convey the same information. The fortune rank and the net income are very small and get lost in the overall viz.
  • There are too many colors. The different colors represent the respective industries as dimensions. They are not used to make a point which, would, for example, highlight the highest profit per second. Using multiple colors for a dimension can be fine, but in this case, it just adds additional complexity.
  • The viz title is formulated as a question but the viz does not guide you to the answer. The big bar chart for Apple get lost in all the information and the immediate answer to the posed question remains uncertain.
  • The viz title only refers to one part of the viz. The title only mentions profit/second all the other information, given in the viz given as additional input, but do no help to answer the questions posed in the title.

My Makeover:

ProfitPerSecond (2)

The key message I want to convey:

  • Apple makes an unbelievable $1.444 profit per second and is the most profitable company in the world. Both of these findings are reflected in the title and the viz by the red color and size of the outer ring.

The design choices I made:

  • I chose a radial bar chart. Mostly, because I just watched a youtube tutorial the other day on how to do it. 🙂 I know that it has a little bit of a bad rep since bar carts would work in most cases just as good or even better because they make the comparison between the different bars easier. Radial bar charts often have distorted proportions and the ration comparison of the different rings are hard. For example, the second ring (JP Morgan Chase) of my viz should be about half of the size of the first ring (Apple). But is this really the case? This is difficult to grasp for the human eye. Even though the radial bar chart holds these downfalls, it is optical more appealing and more likely to grab the attention of the reader.
  • Reduction of colors. Again, I decided to reduce the color palette of the viz and stuck to only grey and red. I could have chosen to color the rings by the industry dimension, just as in the original viz, but I felt that I could convey this additional context just as good in the text box in the middle of the radial chart. Also, additional colors would have drawn away the attention from the most striking finding, namely that Apple makes the most profit/second.

My contribution to MakeOverMonday and my first published MakeOver

Since End of 2017 I take part in the MakeOverMonday Challenge. The essential concept is, to create a makeover of improvable visualizations which have been published on diverse internet mediums. The author of the homepage http://www.makeovermonday.co.uk, Andy Kriebel, provides a new makeover datavisualization beginning of each with week, along with the corresponding dataset.

It is a great opportunity to put various design concept into practice and discover what works for the respective dataset and what does not.My datavisualization tool is tableau. I discovered this program while taking the cousera Data Visualization with Tableau specialization.

From now on I will share my weekly makeovers on my Homepage together with my thoughts on:

  • what I dislike about the original viz
  • the key message I want to convey with my viz
  • the design choices I made to convey this information

I will begin with my makeover of CW2 2018. The original viz is the following:

British both gender rankings-01 (1).png

What I dislike about the viz:

  • There are too many colors. Looking at the viz, no color in particular stands out to me. Red, for example, as a key signal color, is not used at all in the color palette.
  • Size as well, does little to guide the reader to the most important part of the viz. The biggest bar charts are purple and stand for the question: “They have/make a decent amount of money“. On first sight, the size of the balk chart could signal that this attribute is the most important. Only on second sight, do we see that it is ranked sixth. Meaning that in the lowest ranking option it has the highest value.
  • It is tough to compare the values of women and men next to each other. If I want to compare the values of men and women for the respective questions,I always have to move up and down between the 2 separate bar charts and find the corresponding color.
  • The key finding, called out in the title, is not highlighted in the viz. The title mentiones as the most important finding of the data – “Personality is the most important characteristic in a romantic partner, say half of the Brits.” However, the viz does not support this claim visually. Both bars, for men and for women, are just another color in the viz.

My Makeover:

ValueInPartner

The key message I want to convey:

  • I focused on the difference by gender for each question. Since the dataset also provided survey data for multiple countries therefore it was also feasible to focus on the visualization of cultural differences. However, in my experience it can be overwhelming for the audience when the viz tries to answer too many questions at once. I settled for a comprise with the filter option which makes it hard to compare between countries, but allows to switch between them.

The design choices I made:

  • I chose a butterfly chart to facilitate the comparison by gender. For each question, the percentage value by gender are shown along the y-axis.
  • There are only two colors now in the viz. This results in a reduction of the cognitive overload caused by the original viz.