#data fun

[socialscienceside]

26th Oct - Day 24 of the @gradblrchallenge

Not a bad day at all. Graded nine (9) papers, wow. Six more to go tomorrow and then I’m done with 57 of them in total. Then I went home and had an actual, warm, homecooked meal, courtesy of my flatmate who is an angel and far too indulgent of my overworked self. Afterwards, I fixed the mistake in my data analysis I noticed yesterday evening and recalculated all necessary information. Took me only about two hours to do, so I feel like Queen of R right now. Still have some graphics to do, but that’s okay. I’m glad I noticed and decided to actually fix it. Depending on how long the grading takes tomorrow, I might go to a coffee shop afterwards and do it then, or on Friday.

(This is an old picture. I’m a fraud, don’t look at me.)

[socialscienceside]

tfw you're a little wine drunk, listening to the latest fivethirtyeight podcast and just want to put together one of your last missing illustrations and suddenly you notice you've made a small mistake that a) you'll need to figure out how to correct and b) means you'll have to redo quite a bit of your analysis. :))))

[socialscienceside]

22th Oct - Day 20 of the @gradblrchallenge

Day 20, wow to that! I didn’t plan on leaving the couch today, but seeing as I had to retrieve my charger I kind of had to. So since I was back at the coffee shop anyway, I had a giant cup of coffee and a muffin for breakfast and worked on my graphs some more. And then I went back home and alternated between watching Parks & Rec and making more graphics. I got eleven out of twelve done today, so they ended up being fairly quick to make. I actually wish I could show you some, because I think they’re pretty neat, but it might be a bad idea to circulate things from my thesis on the internet before it’s graded and - potentially - published. So I’ll make do with a photo of my screen. ;)

I actually tried to make graphic number 12, too, but that one’s providing me with a bit of a mystery. (Not the graphic itself, but the underlying network partition.) There’s one subgroup in there that consists of several unconnected components and that’s clearly wrong, because unconnected components can not be a group. It’s fairly obvious that the problem is mostly due to one central node that is misgrouped, but I gotta figure out why that happens. So I’ll try retracing my steps and rerunning the algorithm tomorrow and see if I messed up or if it’s just a bad algorithm, lol. Oh well, still lots of progress.

[socialscienceside]

19th Oct - Day 17 of the @gradblrchallenge

Apologies for not posting yesterday. I’m pretty much doing the same thing every day atm and it feels like it’s getting a bit repetitive. Graded more papers today. Also my colleague, who is an actual real-life angel, took some time to show me how to use Gephi. It’s a free software for drawing network graphs and much more intuitively usable than the options in R igraph. So I think I might finish data analysis this weekend? Exciting prospect, certainly! Tomorrow, colleague and I might attend a workshop with some of the people applying for funding for That Research Cluster. Not sure yet, but would be interesting to meet some of the other people involved, certainly.

[socialscienceside]

16th Oct - Day 14 of the @gradblrchallenge

No photo today, because while I did get quite a bit of work done, it was on the couch in my pjs and nobody wants to see that. Anyway, I finally got into data analysis proper and man, let me tell you, it is going well. I’d kind of had the impression already that there weren’t going to be that many road blocks once I got to this point. And it seems blessedly true. I’ve got about two thirds of the numbers I need already. The only major difficulty I think is going to be the plots, as I mentioned yesterday. I’ll try to talk to my colleague about that sometime next week and see if he has any thoughts on the matter.

Working with R has been an experience, to say the least. I keep oscillating between exhilaration when something works and wanting to break the space-time continuum to punch past!me in the face for committing to a project that requires programming in a language I know nothing about. Today was the former category. Particularly when I wrote a piece of code from scratch to loop over all subgroups and calculate their average diameter and it worked on first try. Man, that was something.

[socialscienceside]

15th Oct - Day 13 of the @gradblrchallenge 

Today marks exactly two months to go until thesis deadline. I think I’m doing fairly well. Started the day off slow and slept in, to finally get enough sleep after the insomnia issues of the past week. Think I’ll sleep okay, tonight, too; I’m not feeling so vibrate-y anymore.

Then I got to work on my data analysis again. Most of the day was spent troubleshooting yesterday‘s script, tbh. But I now have all the pre-processing done and tested and all the networks as I’ll need them. Will start on the subgroup analyses tomorrow. This evening, I made some drafts for tables in Word, so I can start inserting the relevant information as I go along tomorrow. I also played with the plot options in R igraph a bit, but I’m not too happy with them, to be honest. The graphs igraph creates are really crowded and not very legible. I’d like something that arranges nodes according to community membership, but as far as I can tell there’s not an easy option for that. More research required on that front.

In terms of self-care, I also took a lovely walk through the park. The leaves are just starting to turn and it was really quiet and calm. Things that also happened included groceries, laundry, and making myself a delicious veggie burger.

[socialscienceside]

8th Oct - Day 06 of the @gradblrchallenge

This is really short and boring today, but. Wrestled R all afternoon. It looks like the things I assumed were going to be tricky are actually fairly straight forward. Meanwhile things that should be super easy, such as dichotomizing a network matrix at different threshold values, are making me want to tear each hair on my head out individually, so that’s fun. Managed to not throw my computer out of the window and I think I mostly know what to do now, although I was too braindead by the end to actually try it.

More on that front tomorrow. I also want to start going over some notes I took during my reading period and didn’t yet incorporate into the writing process. So not quite out of work yet. Hope you’re all doing well!

[socialscienceside]

Not to drag anyone or anything, but @ R package developers, are you sure that factanal was the best shorthand for factor analysis you could’ve come up with?