I have a chart like this: Chart

There are some problems that I am stuck:

  1. Some columns value are very small compared to others, so I cannot identify whether they are zeros or not (I used logarithm scale, but even with log10 scale the problem still remains and I don't want to increase the logarithm base because I might have to do some calculation in my head to know the columns values)

  2. I cannot easily see the trend of yeartodate -> month -> week -> today data without comparing them one by one in my head. I usually use 4 columns but for this chart there is too much data so I have to combine them in 1 row to save space.

I wonder if there is a better way to visualize these data. Is it not necessarily a bar chart.


1 Answer 1


I think it's quite good but it represents a lot of information at once. There are 4 dimensions, which is a bit too much:

  • proprietary vs foreign
  • names on Y axis
  • time unit
  • value

Sometimes it's better to simplify the dimensions and present different graphs which give different perspectives on the data.

The time unit is indeed the least readable here. Ideally it should be represented consistently, for example with 4 different bars for every name. However I can see that this would make the graph too large given the number of groups on the Y axis.

Another idea would be to represent the value with different colours, typically a gradient. This would free the X axis which could be used to represent the time unit dimension. The graph would look like a grid (two grids for proprietary/foreign) where each cell represents the value with gradient/colour and the time unit are more easily readable/comparable. You would lose the precision of the value, but if the goal is to observe the most important trends it shouldn't be a serious problem.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.