I’m having problems with this Titanic data set.
The problem is mainly how to tell a compelling story ?
I know my variables of interest well, those include Pclass, Age, Survived, Sex. I’ve spent quite some time exploring the data set (exploratory graphs) and found that a mosaic plot explores the trends in the data set really well. Here’s a picture I found on r-bloggers showing the mosaic plot
So I guess I’m good with the exploratory graphs, but I really have no clue about the explanatory ones. How do I tell a story about my findings of the data set that has interactions and animations ?
Even if I found a way to make this mosaic plot interactive, users won’t probably understand it from the first glance because it is a bunch of rectangles with different sizes.
I tried a couple of data visualization tools like RAW and Tableau, but i didn’t find a graph that would show the trends in the data in a simple way and is intractable and has animations.
Thanks
1. For your mosaic, I’d say strip the residual info and collapse it down to a histogram/mosaic of yes/no survival. This should be more readable. I’ve included in the top of the attached image and I find it more readable.
2. For explanatory graphics, I’ve always been partial to a partition/tree for this data-set. It let’s you see what the important factors are rank order and highlights some of the more horrifying aspects (3rd class children never makes for comfortable reading). I’ve included an example in the bottom of the attached image.
3. For the interactive part of the story, why not ask the reader to select tickets for them and their family and let them know the chance of their survival. You could do it as a chance of survival or have a randomly selected trip from the probabilities and report who survived or not. I’d guess everyone would click more than once. Maybe have a % chance of survival show up after a few clicks.
Look at the statistics for the crew. A very different story.
You could use the percentage of people that survived instead of dead and survived.
I would leave the residual info out and use pink/blue for the sex.