Water Resources Discipline ITAC Activities

Jessica,

I'm inclined to agree with Ginny (and Joyce too?) that a our audience is better served, particularly in the long run, by putting our limited resources into improving and expanding the Help for the USGS library. In a few Help pages, we can address a particular topic in the detail needed without worrying about overloading the user in a short (1/2 -1 hour) Cyber presentation and the instructions are then inside of S-PLUS for all future users. And, by way of the examples we include, the hands-on practice is provided.

Dan, I recommend that, as you prepare to answer Jessica's question, you review the Help provided with the USGS library for S-PLUS as Ginny and Joyce suggest and give us your feed back on what you find - and don't find - there.

Also, did you take the Basic Hydrologic Data Analysis Using S-PLUS class when Kathy Neitzert and Cherie Miller gave it in PA in March, 2002? As Joyce points out, there have been major improvements in S-PLUS and the USGS library since then, but at least it's a start.

Regards,
Jim

----------------
Jim Slack                  Internet: jrslack@usgs.gov
U.S. Geological Survey
415 National Center        Voice: 1-703-648-5714
Reston, VA  20192          Fax:   1-703-648-5722

Virginia, Joyce,

When I mentioned cyberseminar I was thinking a session no more than 30 minutes long where you would address for example "boxplots" just one type of graphic. I think the scientists could join the cybersem right from their computers. You could even allow them a few minutes to switch from the Cyberseminar window to the Splus window so that they could try it out "hands-on". Might get a bit complicated, I know. I guess the other thing scientists can do is follow your instructions and call one of the Splus people when they get stuck.

Dan, do you think short cyber seminars, focused on a specific topics might help or is the second approach better?

Jessie

P.S. Or you could even remote control their desktops so that you can walk them through a process...but this would require a slight learning curve for the scientist too...hmm, let me see what else I can come up with.

Dan and Jessie,

Additional resources on graphing are in Splus's USGS help >Help>Available Help>USGS, Contents: USGS Tip Sheets, especially "Workflow for modifying S-PLUS graphs for legibility and USGS style" (by Wendy Danchuk) and "How to create graphs".

Joyce,

As far as a cyperseminar, I also need to think a bit on this. I wonder if what is needed is more examples of creating and modifying graphs in the Tip Sheets -- and the cyperseminar could go over a subset of the examples. Maybe we also add to the Help -- "things Splus GUI cannot do, cannot do well, or cannot do easily"--example embedding data in plots.

Or maybe, what is needed is an intro to Splus 6.1 with USGS enhancements as it is--just go over a shortened version of the class--so people know what is there. I have mixed feelings about whether a cyperseminar is the best media for disseminating this type of information--it may be too much information for a 2 hour session. I would be interested in what the rest of the Splus Help group thinks.

Ginny

Virginia (Ginny) L. McGuire
U.S. Geological Survey
100 Centennial Mall North, Room 406
Lincoln, NE 68508

vlmcguir@usgs.gov (402)437-5124

Dan and Jessie,

I have included some comments within Dan's questions and concerns below. I would have had a tendency to agree with Dan on many of his comments prior to the last two versions of Splus, with the latest (6.1) being a huge improvement; not only in what we can do and the ease in doing it, but also with the online help. Slowly but surely I think we are getting Splus to where it needs to be.

I will start by saying that the S-Plus group does offer to teach classes at the district (see http://wwwqvarsa.er.usgs.gov/splus/tipsheets/splusgraph.pdf There are also a number of online documents designed to simplify the process of going from the default Splus plot to a USGS type plot. One that is very good is located at: http://wwwqvarsa.er.usgs.gov/splus/USGS_Supplement.pdf

I hope this helps. I know these major changes are not easy simple because of the learning curve which cuts into our productivity, especially when there isn't enough time in the day as it is.

Jessie,

As for a cyber seminar, I will have to think some more on that. Virginia McGuire is really excellent at taking S-Plus standard graphics to USGS-style graphics. I guess if I were doing one, that is what I would focus on. I am including her in this response because she may have a better idea of how much time that would take, etc. I do feel that much of the frustration with Splus is exactly that - getting the plots to look like we want them too without our cartographers having to clean up a whole bunch of things.

Joyce E. Williamson
(605) 355-4560 x219
jewillia@usgs.gov

Joyce,

Could you address Dan's concerns below and maybe start a list of topics that the field needs help with and can probably be addressed via a topic specific cyberseminar?

Thanks,

Jessie

HI Jessie. I had not given it alot of thought lately. We did get a new version of splus installed her within the past month with the environmental stats package. I really have not had the opportunity to use it. I know that the env. stats package can do probability plots, which raw splus can not do. So, that was one thing that g2 could do that splus could not. but that is covered now.

My main concerns with splus are the formatting of plots (I use SAS for stats, so I'm unclear on that) IT is not very straightforward in applying a set number of minor ticks between the majors. This isn.t too bad once you have done one or two. There is a USGS help sheet called Workflow for modifying S-Plus graphs for legibility and USGS style that steps through taking a default S-Plus graph and making it fit the USGS style.

It is difficult (probably I just don't know how to do it correctly) to get multiple plots on a page. You can do it, but there appears to be problems with frames overlapping for the multiple plots.

My first try at this was not very productive either until I read the online help. There are multiple ways to do this, but I will include one. Others are detailed within the online help.

Adding a Graph to an Existing Graph Sheet

To add a graph using the Insert - Graph option

1. Open a Data window containing the data to plot, or view it in the Object Explorer. Select the data columns you want to plot.
2. Choose Insert - Graph.
3. Specify the desired graph and plot from the Graph Type and Plot Type lists.
4. Specify the name of the Graph sheet to which you wish to add a new graph from the Graph Sheet list. Choose OK to close the dialog. The graph appears.

To arrange them:

1. Choose Format - Arrange Graphs.
2. From the Arrange Graphs submenu, choose the desired layout style: Default Size/Position, One Across, Two Across, Three Across, Four Across, or Overlaid. This determines the number of graphs to place horizontally before "wrapping" graphs to the next row.

To specify the order in which the graphs are laid out you can select them in the desired order before using Arrange Graphs. Use Shift-click to select multiple graphs and then choose a graph layout. Be sure to select the entire graph area when selecting graphs. You can use Arrange Graphs as many times as you wish to experiment with different layouts and to rearrange the graph order.

If you round the axis number - say you have 6.8 and you round to 7, the data are not shifted. That is, if the data point equals 6.8, and you round the axis to seven, the data point will appear to be seven, so it actually rounds the axis and the data.

This is what I would expect a graphics package to do. The minimum has not changed (still 6.8), only how you are presenting it has changed . you are rounding things which includes the data as well as the axis. If what you really want is an axis from 7 to whatever and you want to truncate the values below 7, then you need to specify a axis range with a minimum of 7 and then the 6.8 value is not displayed, no matter what rounding is set to.

Using a date formatted variable on any axis is difficult to format to get any readable date labels on the axis. This has been a long-standing issue with S-Plus and has improved dramatically with each new revision. Working with dates in version 6.1 is pretty straightforward and a number of built-in USGS help pages have been written to help users get exactly what they are after for time-series data.

G2 can read card-image data from adaps; splus can not - but splus can read rdb files from adaps, so I guess that really does not matter. Correct, RDB is the standard for exporting/importing from ADAPS to S-Plus

G2 can color or shade differences between two plots on the same graph. I don't think splus can do this, but I'm not sure. S-Plus has been able to do this for some time. To create a shade-between graph, select Area Chart from the Graph, 2-D Plot menu or the Area icon using the 2-D graphics icon toolbar, selecting the dataset and variables in the standard graph menu. Under the Options tab, select Individual for the Y-Values option instead of the default of additive. The fill direction is also selectable from this menu. Fill properties can be set initially in the graph menu or later. The order that the Y variables are entered determines the order of the shading with the first variable being in the front. Y variables can be entered either manually, separated by commas, or by selecting through the menu, using control-click for each selection after the initial selection. boxplots - I have done boxplots with splus (not in the usgs boxplot module). The problem I noticed was the inability to change the whiskers and outlier ranges. I am unclear on this right now, but I know that the old accepted version of a USGS boxplot could not be produced by the boxplot option in splus. When I tried to use the USGS boxplot module, that was not working in our district, so i am not familiar with that. I spoke with our reports unit, and the splus boxplots (non usgs) are acceptable, they just have to be defined correctly. When I heard that, the boxpots were not a concern to me any longer. The USGS boxplot option should allow you to create a correct boxplot including the standard whisker and outlier ranges.

There's probably a myriad of other things that I have not had to deal with, but overall, I think the graphs coming out of G2 are of higher quality than splus. I have not met anyone who used both who did not agree with this. And we know the survey is very "uptight" about the graphics in the reports produced. That aside, splus is much better for exploratory data analysis and cranking out numerous plots very quickly so that you can see what is going on. G2 is not for that type of thing.

When G2 first came out, I recall hearing a lot of the same complaints that I now hear about Splus. It took quite q few years for G2 to be as powerful as it was in the end. The problem with this custom built software is that once the person who wrote it left the USGS, we were really in a bind to support or add to it. It was also written such that it would not run on something other than the DGs. The reality is that it would have been very time consuming to port G2 to the Sun or desktop environment and something had to be done on a faster time scale. For those offices and employees who didn.t use SAS, a statistics package was another major issue. S-Plus could accommodate both and is written in such a way that custom graphs and statistics could be added over time. It has been a couple of years but at least in our office, everyone has made the switch pretty easy once they got in and started using it.

THe big plus for splus over G2 is that you can subset in splus, so basically you can do plots or stats on different subsets of the same data set. FOr G2, the data have to be subsetted prior to bring the data into G2. That is very cumbersome and time consuming.

Lastly, I think when you were at our office for that meeting, our district had splus on a relatively poor quality server. I say that because after the meeting, splus was moved to another server and alot of problems or aggravations were solved. I had many problems with cursor to screen interactions, and it appeared to be some sort of memory problem with the server (someone told me that, I have no idea). Splus would also just boot me out, delete graphics, not allow you to undo, or if you undid, it would not really totally undo it (the graph would be moved from it's original position, stuff like that). Anyway,after the switch to the new server, I did not have the interface problem or the other irritations. The switch to the new server was very important. Splus was almost unusable on the old server(s). I'm not sure of the differences between the servers.

I did not talk with Jim, so I guess you got to hear all the ranting. I appreciate the fact that you have followed up on this. I'm sure you got a ton of other things to be concerned about.

dan

Hi Dan,

I'm at an ITAC mtg...found someone that might be able to better address your GS2-Splus related questions...do you still have questions/issues that you feel SPLUS is not addressing?

Joyce Williamson is in this office, SD, she used to be a G2 user and thinks that her needs are now met by SPLUS. She is also part of the SPLUS-team. She says the current release does box plots and other stuff I am not familiar with. If you have a chance tomorrow or Thursday, could you briefly list what the outstanding issues are so that I can forward to her and we could all understand what still needs to be done?

Thanks,
Jessie

Hi Jessica,

No, I haven't; I've been swamped in general. I'm not sure what you mean by the support question; DIS has been only part of the support for S-PLUS. Let's talk when you're in next.

Jim

----------------
Jim Slack                  Internet: jrslack@usgs.gov
U.S. Geological Survey
415 National Center        Voice: 1-703-648-5714
Reston, VA  20192          Fax:   1-703-648-5722

Jim,

Have you had a chance to call or email Dan regarding this issue? With all the re-org chances in DIS do you still feel SPLUS has appropriate support for the field? We definitely need it... people are still using G2, we can't stop relying on the DGs until they stop using G2.

I'm working from home today, else I'd pay you a visit.

Thanks,

Jessie

Hi Jim,

I have an action item from the PA ITAC meeting back in June, to get you and Dan Galeone, talking. You had mentioned that you did not have access to or ever used G2 and in order to understand the functionalities hydrologist wished SPLUS had you would need to work closely with someone that knew G2.

Dan Galeone, volunteered (or was volunteered) to help you understand the functionality hydrologist wish G2 had.

Could you please contact him, so we can opened up the channels of communication?

Thank you,

Jessie