I did a little work with D3 to build an interactive 3D scatterplot but I remain skeptical of the value of a lot of the interactive data displays I see. Looking at the first example you point to - http://bl.ocks.org/diethardsteiner/3287802 - I can't help but wonder if a static representation like a group of barcharts is more useful. It's fun the way the barchart on the right changes when you select a segment in the pie chart but notice that the highest bar for "Steve", representing 3,443 apples, is the same as the height of the 25,922 apples for Sam. Of course, since you can't see the charts for both Steve and Sam at the same time, the distortion of having the same height represent different numbers is hidden.
Similarly, this chart - https://observablehq.com/@d3/countries-by-area - of the areas of different countries is not as useful as a simple table of countries and their areas since you have to hover over each picture of each country to find out which one it is. It's fun and all to watch them light up and display information when you hover over each one but a simple table with all the information available at once would be more useful for most purposes. I have noticed this drawback with a lot of the interactive data displays I have seen: you can't look at several items at the same time because you have to hover over each one to display data for that item only. I see a lot of this enchantment with novelty over-riding practical concerns of usage. Recently when my wife showed me an alluvial (or Sankey) diagram proposed for a publication on which she works, I found it hard to draw many useful conclusions from it as it was one of these with multiple layers - see these http://www.datasmith.org/2020/05/02/alluvial-plots-vs-sankey-diagrams/ for examples that are perhaps overly complex. These types of diagrams are also sensitive to the order in which the data categories are laid out. When I briefly researched alluvial diagrams, many of the hits were for different software packages for creating them but there was no evident critical analysis of them outlining what they are good for, their drawbacks, or guidelines for how best to construct them. On Sat, Nov 7, 2020 at 1:46 PM Brian Schott <[email protected]> wrote: > Before I retired, statistically interactive displays intrigued me. I became > aware of software by Datadesk which facilitated looking at data in the way > that the current dashboards suggest; at least the similarity is suggested > in my mind. > > So my reason for posting this message is to see if others have been > successful in using J in this interactive visual manner for statistical > data. My research on this subject has mostly centered on D3 and I have > included some links below suggesting what has informed my research the > most. I have been rather disappointed in the complexity of programming D3, > and look forward to learning other's experience. > > The very first link shows a running dashboard and related JavaScript code. > The second and third links provide code that is embedded in an article > written in the statistical programming language SAS; this link is included > because it might suggest how D3 can be linked to other programming > languages. (By the way I have seen the demos for D3 in both JHS and jqt.) > The fourth link promotes the possibility of using SVG for three-dimensional > data. Apparently that has been a weakness for D3. > The fifth link is to a rather long discussion about the changes in D3 which > are being pushed by the original developer. Many people are arguing against > the changes because they move D3 from being a library to a separate new > application. > > http://bl.ocks.org/diethardsteiner/3287802 > https://gist.github.com/NPashaP/96447623ef4d342ee09b > https://www.mwsug.org/proceedings/2014/DV/MWSUG-2014-DV09.pdf > https://observablehq.com/@sanderevers/svg-transformations > > https://talk.observablehq.com/t/i-want-to-learn-d3-i-don-t-want-to-learn-observable-is-that-ok/1957/6 > > Thanks, > > (B=) > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > -- Devon McCormick, CFA Quantitative Consultant ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
