How about using areas rather than lengths, so either a pie chart or a histogram? Also do you really need to show the total value as a bar? What if you just showed the dollar figure and a % of the total, so just the red bars and some text $486,007,475 out of $39,592,226,311 (1.25%).
I would look at exactly what metric you are showing and think about whether it is the right one. Do people really understand 39 billion, 592 million, 226 thousand, 311 dollars? 1.25% may feel more meaningful and tangible (although it may not be the message you want to convey). On a separate note, there is a general tendency to be misleadingly accurate with numbers. It must have been some calculation to get that figure to the nearest dollar. It may be better to say $486M and $39.5Bn. My (skeptical) guess is that that the number has a large degree of error attached to it and it is +/- a few hundred million. In my experience few people in the general population understand log scales and don't realise a 5 earthquake is 10 times as strong as a 4. Simon --- In flexcoders@yahoogroups.com, "charlie.szymanski" <[EMAIL PROTECTED]> wrote: > > I have an interesting issue with a mini charting application I am > trying to create. > > You can see my current attempt here: > http://ibiseye.com/?stormID=AT200403&active=1&season=2004&wind=145&name=Charley&zoom=5&lat=26.3&lng=-71.05000000000001 > > and click "Charley Synopsis" > > Basically, we are using hurricane wind-fields and a giant database of > all the property parcels in the state of Florida to calculate how much > property value was "at risk" during a given hurricane (this is live > during the season if a storm is forecasted or does go over Florida). > > The problem is that the numbers are exceptionally varying in range. If > you look at the bars on that Flex Chart you'll notice we are using the > Logarithmic scale . The reason we had to do this is that the "all" > data (or even the data within the different categories if the 'all' > data is removed) is usually hugely different in value. This difference > is exasperated by differences in property types. What happens is you > end up with a chart where the bars are barely visible for a number of > the property types (which are even less visible if the chart is not > scaled logarithmically for the "all" category). > > If you mouse over the current chart values, you'll get the actual > dollar value of each bar. It's easy to see (especially in the > Residential type) just how misleading the chart could be. They look > almost the same, but the blue bar is actually 5x greater than the red one. > > Has anyone run into this issue before and do you all have any good > ideas of how to get around it? Right now I'm not really even looking > for code, but more of any creative solutions to displaying that > information in a way that isn't misleading (which the current scale > may be -- you have to remember how many old people live in Florida). > > Thanks for any insight, > > Charlie >
