psyteachr / msc-data-skills Goto Github PK

At least on Ubuntu, this line in exercise() does nothing.

Consider using rstudioapi::navigateToFile(filename) which will open the file in Rstudio.

To allow for anyone not using Rstudio, it might be worth either using usethis::edit_file directly or copying the last few lines from that function into exercise() which tests if Rstudio is being used and opens the file in an appropriate way.

I can send a pull request if you want, but cannot test on windows machine

I cannot get the answer for question 4b to reproduce, using

set.seed(321)
mu <- seq(0, 1, 0.1)
samples <- map(mu, rnorm, n = 100)
sample_means <- map_dbl(samples, mean) %>% print()

produces:
[1] 0.009067669 0.098448711 0.363801400 0.384951951 0.503663364 0.346219426 0.648143250 [8] 0.805276960 0.838016778 0.892675777 1.051627663

maybe an issue with set.seed() working differently?? Not sure how the rnorm function interprets mu to contain mean values in this case as opposed to say SD values?

The CSV used in 08_sim_exercise.Rmd at Question 7 is US CDC Growth Chart Data Tables.

This CSV file is confusing in a few ways. First, beginning in line 435, it looks like this:

1,240.5,162.4776756,166.0916092,169.6925342,173.2808172,176.8568057,180.4208299,183.9732035,187.5142253,191.0441799
,,,,,,,,,,
,,,,,,,,,,
,,,,,,,,,,
Sex,Agemos,-2,-1.5,-1,-0.5,0,0.5,1,1.5,2
2,24,78.02185636,79.76434516,81.50408026,83.24112888,84.97555512,86.70742016,88.43678251,90.16369812,91.88822059
2,24.5,78.40361515,80.15493488,81.90429469,83.65174068,85.3973169,87.14106545,88.8830266,90.62323896,92.36173951

Also, Agemos has range from 24 to 240.5. What is the Agemos?

You can get different answers depending how you calculate as one individual in users and users2 shares a user_id "16498" but has a different gender and birthdate in each

In 08_sim_exercise.Rmd on question 4, a power-level is to be estimated through simulation. The required simulation level before getting a correct-flag is 10000 (or 1e4).

I got

Error in loadNamespace(i, c(lib.loc, .libPaths()), versionCheck = vI[[i]]) : there is no package called 'tidyverse' Calls: <Anonymous> ... loadNamespace -> withRestarts -> withOneRestart -> doWithOneRestart Execution halted ERROR: lazy loading failed for package 'dataskills'

when trying to install had to install tidyverse manually as it does not seem to be registered as a dependency.
Also this contains some packages that need compiling in windows so you need Rtools40 setup
Also watch out for Microsoft R Open and its shitty CRAN repository snapshot
Was hoping R would be less of a shitshow than python with package management alas

In 08_sim_exercise.Rmd on Question 7:

Calculate power for a two-tailed t-test with an alpha of .005 [...]

Instead of alpha, one could use significance level. There is confusion between confidence level and significance level, so using both in the exercises is really good.

When compiling this, I noted that it would be useful to put into an R package so that package dependencies could be installed automatically and the entire book could be installed locally with devtools. Would this be useful for me to send over as a pull request?