Comments77

From InterSciWiki
Jump to: navigation, search


Ready for the first day of Xiamen classes

Today I spoke to Ren Feng for an hour on a China International Phone card and gave him visual introduction exercises at the intersciwiki for his students.His class at Xiamen University (Amoy=Xiamen), Fujian province starts Monday and goes for 15 weeks. After we finished Tom Uram called to say that he had simplified version http://SocSciCompute.ss.uci.edu to four options: EA LRB SCCS and WNAI. We walked through model construction and results for maps and for DEf regression. Tom is now implementing version 4, DEf01b and the two mapping routines, Anthon's black autcorrmap and Uram-Oztan-White's color Rworldmap.


Here is the Introduction I gave to Ren Feng: At the intersciwiki Main Page - click Visual Manual. Follow the logic of the independent variable and its predictors and how the colored maps are created. Copy the independent variable list, click the SocSciCompute window, click SCCS, enter the depvar, then the independent variables, comma separated. Click the blue EXECUTE button. Then click 3: SCCS map, save (click Diskette image), open at your computer, and print. Then click 1: SCCS h, save (click Diskette image), save, open the *.csv file, fill in empty right-hand cells 1-10 for the h functions to match the documentation, and discuss the content.

When the 60 minutes cut off, we were ready (tomorrow?) to finish exploration. Trying out different models after the phone call, the SocSciCompute site went down completely. I was pleased to see, about 10 minutes later (our possibly due to help from our SocSci computer lab), the site was working again. I'll ask if it was them or self-recovery. Anyway, recovery from a crash is a good sign for classroom usage.

Next session: click the R gui Examples at the top of the Visual Manual page, and there, at the new Examples page, click Go directly to CoSSci Models: which give R gui models listed with distance and language autocorrelation effects.

Ten of the models listed from the link Go directly to CoSSci Models (which connects to List of Rgui to CoSSci Models) have (SCCS) RiV marked in bold, each of which has a dpV and a set of RiV independent variables. You can experiment entering some of these (only those with v###, ### for a number) in the first two windows at http://SocSciCompute.ss.uci.edu. Again, with R gui, you can ask for maps for some of the variables you have entered. It is possible, at the end of a run, to change the .FALSE. to .TRUE. in the h<- doOLS function in order to compare distance and language autocorrelation effects alone with distance, language and ecological autocorrelation effects: i.e., to run each option in sequence

#activate ecological autocorrelation with  ew = TRUE
h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = oxog, dw = TRUE,
   lw = TRUE, ew = TRUE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE,
   slmtests = FALSE, haustest = c("v213.d3"), mean.data = TRUE, doboot = 500)
CSVwrite(h, "olsresults.ew", FALSE)
#deactivate ecological autocorrelation with  ew = FALSE
h <- doOLS(smi, depvar = dpV, indpv = UiV, rindpv = RiV, othexog = oxog, dw = TRUE,
   lw = TRUE, ew = FALSE, stepW = TRUE, boxcox = FALSE, getismat = FALSE, relimp = TRUE,
   slmtests = FALSE, haustest = c("v213.d3"), mean.data = TRUE, doboot = 500)
CSVwrite(h, "olsresultsNo.ew", FALSE)

Tom Uram

I've separated DEF2a into these four tools, each of which has a working set of default values, and points to Anthon's web page for help (via the link at the bottom of the page).

I've also included a copy of these tools in a section titled "HPC Tools"; these tools will run on Trestles.

All other tools have been moved into the Development Tools section.

I removed several more bio-specific Galaxy tools to simplify the menu for the students.

I may make further improvements here, but will try to leave things mostly the same to avoid making many changes right before the start of the class. I'll also try to integrate the other mapping features as we discussed on the phone today. I'll let you know about progress on these points.

Tom

Ren Feng

Doug:

ok, I got the outline of the new approach under your guide. I'll keep exploring it. Is the wiki page still available for the students to post their results on it?

I plan to start the class with the background of cultural anthropolgy and sccs database. So those students may go to the analytical process four weeks later. Because they come from different departments and most of them have not basic sense of social science and statistics.

Best

Ren Feng