In which I fight the good fight against R
Sep. 18th, 2021 08:34 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
redfiona99(Mostly for ioplokon who saw the questions but not the reason behind them)
So, partly in follow up to some ideas from my 2019 Rugby Union data visualisations (https://redfiona99.dreamwidth.org/1193366.html) and partly in advance of next year's StatsCup, because I have grand plans, I wanted to find a way of mapping players on a team that play together as a unit.
This is somewhat complicated by my complete lack of coding.
As you know, I occasionally link to Rstats posts; this is mostly so I have links to things that I can crib code from.
Most recently, I read this marvellous post by Jim Vallandingham - https://vallandingham.me/shaw_bros_analysis.html Now, I really like this post because it gives code and a link to his data so I can figure out the format my data needs to be in in order to replicate the analysis. If more people could do that, I'd be grateful.
If you read the post, you'll see it contains a way of looking at actors who co-occur. And I thought, ding, ding, I can warp and spindle this to fulfill my aims.
I began with the creatives from the films in my 100 Things set of posts (https://redfiona99.dreamwidth.org/tag/100+things):
1 - 2004 is over-represented in my list of "100 Great Scenes in Not So Great Films"
2 - The following people are involved in more than 1 film:
3 - If I compare the director, writers, producers, editors, cinematographers, composers and first 10 actors from the first 10 films, I get a heat map like this:
4 - Looking at only those who are involved in more than 1 film, the heat map looks like this:
I told you there was a "bad Marvel film" cluster, and this bears it out.
Unfortunately, as of yet, there aren't enough to let the network diagram be coded (the code does work because I have since run it on some rugby data from the 6 Nations).
Other things I have learned:
1- I need to update R and RStudio more regularly than once a year
2 - library (package = "name") is my friend
3 - capital letters are important in R
ioplokon who saw the questions but not the reason behind them)So, partly in follow up to some ideas from my 2019 Rugby Union data visualisations (https://redfiona99.dreamwidth.org/1193366.html) and partly in advance of next year's StatsCup, because I have grand plans, I wanted to find a way of mapping players on a team that play together as a unit.
This is somewhat complicated by my complete lack of coding.
As you know, I occasionally link to Rstats posts; this is mostly so I have links to things that I can crib code from.
Most recently, I read this marvellous post by Jim Vallandingham - https://vallandingham.me/shaw_bros_analysis.html Now, I really like this post because it gives code and a link to his data so I can figure out the format my data needs to be in in order to replicate the analysis. If more people could do that, I'd be grateful.
If you read the post, you'll see it contains a way of looking at actors who co-occur. And I thought, ding, ding, I can warp and spindle this to fulfill my aims.
I began with the creatives from the films in my 100 Things set of posts (https://redfiona99.dreamwidth.org/tag/100+things):
1 - 2004 is over-represented in my list of "100 Great Scenes in Not So Great Films"
2 - The following people are involved in more than 1 film:
3 - If I compare the director, writers, producers, editors, cinematographers, composers and first 10 actors from the first 10 films, I get a heat map like this:
4 - Looking at only those who are involved in more than 1 film, the heat map looks like this:
I told you there was a "bad Marvel film" cluster, and this bears it out.
Unfortunately, as of yet, there aren't enough to let the network diagram be coded (the code does work because I have since run it on some rugby data from the 6 Nations).
Other things I have learned:
1- I need to update R and RStudio more regularly than once a year
2 - library (package = "name") is my friend
3 - capital letters are important in R
