2011/3/29 Ammar Yasir ammaryasir.88@gmail.com:
On Tue, Mar 29, 2011 at 5:01 PM, Marc Delisle marc@infomarc.info wrote:
Ammar Yasir a écrit :
Thanks
I had a doubt (it might be silly but I thought I should ask). One important feature of such a data visualization system is that it should not create confusion or represent something that the data is not. Consider the movie database example, when a user sees a plot of ratings vs year / genre vs lanuage, a lot of data points will overlap. So a lot of data will be hidden from the user at the first sight. What will a user make out of it? I know we handle such conditions by providing all the data instances when he clicks on the point but a user might not go that far. If he is not able to understand the initial plot he might not continue further ( like not zoom in/click data point). I am just a little concerned about the usability part and not the functionality.
I don't think that the plot contains all that. In the example I have seen, the X axis contained "year of release" and the Y axis was for "popularity level from 1 to 9".
No I am saying that a lot of movies will have same year of release and same popularity. Those movie points will overlap on the 2D space, what then?
One solution is to allow for adding random jitter to point coordinates in order to scatter them around their real position, like it's done for scatter plots in RapidMiner ([1], last screenshot).
[1] http://rapid-i.com/content/view/9/199/