At last, Sunbelt XXVII begins - well, tomorrow, actually. Before the conference proper there are two workshop slots. I'm attending two Pajek sessions. It's one of those tools that has a lot of power, if you can just get your head around the way it thinks. As much as anything, I'm looking for the little process shortcuts that will make the analysis process more manageable. So far...
- New feature: using partitions to define start and end-points in temporal networks.
- The functionality hidden Layers menu is much under-valued - its a great way to visualise hierarchy, and offers a few different options for minimising line-crossings.
- Macros are useful both (i) for automating repetitive tasks and (ii) building complex functions from Pajek's panoply of parts (when you've got your head around the latter!), but take care that you have the right network, partition and vector selected in the main screen before you run them.
- Quote: "Pajek is much more a program for exploratory SNA than confirmatory SNA" [emphasis added].
- Excel and Access are both effective data storage options - an Excel course is a must, to get the best out of it in this role.
- You can deactivate the warnings that Excel makes every time you try and save in TXT or CSV format (thanks, Jurgen!).
- Excel's Data/Text to Columns... function allows you to do the same conversion between text and columnar data that you get when you open a txt file directly into Excel. Did everyone else know this?!
- The Excel-to-Pajek software ignores weightings, but Text-to-Pajek (available here) can handle link values and much more.
- There are two standard values for signaling missing data within Pajek - 9,999,998 and 9,999,999. And it does use them - for example, in frequency calculations.
- Moving data from R to Pajek is easier than from SPSS to Pajek (from Pajek to either is easy). R looks more and more interesting, in fact.
And that's it , so far. More tomorrow, when there also will be papers to report. 'Til then.