Hi Akshaya, I'd spend some time thinking about exactly what you are trying to show the users. Looking back at your original graph, I have several questions that come to mind:
1. Are the "I" and "V" categories really showing something that should be on the same graph together? It looks like one is "attendance problems" and one is "reasons why employee left companies". If so, it would probably be much better to split these into separate graphs, one for attendance problems and one for reasons why employees left. If not, they should all be on a single graph, maybe (?) using bar color to show the two categories.
2. When you say there are "many more reasons" - high level how many different reasons are there? 10 vs. 100 vs. 1000 will drastically change the visual you may choose (but maybe not...bar graphs probably still best)
3. I am not a huge fan (at all) of tree maps. Data in them is very difficult to understand without seeing the underlying numbers - at which point you might as well just show a sorted list. For example, spending a few minutes looking at the treemap, the questions I had were:
- Is violated rules 1 > disssatisfaction with pay 1?
- Is violated rules 1+ attendance issues 3 > failure to report 4?
- I looks to be larger than V - but by how much?
4. The short answer - what EXACTLY is the question you are looking for this to answer? Counts of total I vs. total V (show just the two numbers). Proportions of I vs. V, broken down by subcategory (use separate I vs. V bar charts with one bar per reason). All reasons I&V together with the most common of both (one chart, use bar color to differentiate the two).
It's a nuanced question...you'll need to decide what the question is and what best answers it. IMO treemaps not a great choice because they depend on the ability to compare "areas" (which humans suck at) vs. comparing lengths, which we're great at.
It's an interesting conversation - let us know your thoughts!
Scott
