Taking plot() to the limit one more time on line 114, the code breaks (no output). Here's a clunky solution: make a new column as.factor to pass through the col and pch argument.

gap$plot_cat <- as.factor(gap$continent)

scatter <- plot(x = x, y = y,
main = "Life expectancy versus gdpPercap",
xlab = "Life expectancy (years)",
ylab = "gdpPercap (USD)",
# Change point colors to correspond to continents
col = as.integer(gap$plot_cat),
# Change point symbols to correspond to continents
pch = as.integer(gap$plot_cat),
# Change point size
cex = 2,
las = 1)

I don't know whether we should recommend the psych package or dplyr for summarization in PART 3-2. I understand that we need to teach enough base R but psych is not base R and its syntax is quite similar to dplyr and we introduce ggplot2 soon (PART 3-4) anyway.

Part2.R - Line 161
install.packages(dplyr)

Needs to be changed to
install.packages("dplyr")

Move

• install.packages()
• library()
• ?
• mean()

to more appropriate sections.

Instead, let's lead with a summary of what the four window panes do in RStudio.

Currently, challenge 4 has the users develop plots based on the data frames. This is problematic since we haven't covered any plotting. It seems a bit out of place - maybe this should be a challenge about merging?

Currently the ordering is:
functions > for loops > if-else statements > functions again

Proposed new order:
if-else statements > for loops > functions

Notes:
if statements are very intuitive and many have used them in programs like Excel, then for loops are super important and second-most intuitive, then functions can combine all this together

Here's penguine dataset: https://github.com/allisonhorst/palmerpenguins

Also, the iris dataset originated from the Annals of Eugenics, in 1936, published by Ronald Fisher. It's time to reconsider its usage in the intro stat and data science courses.

There were wonderful updates to the content for Part 2.R but the corresponding solutions file does not appear to be fully updated (see merge section, maybe others)

There should be a short example demonstrating that = can be used for assignment in R, but <- is preferable due to style (and that in specific examples that don't need to be detailed at the time, they do in fact behave differently).

The README needs a few updates to bring it up to style. It should follow the template set by R-Data-Visualization.

We should include some information about the datasets, either in the README or in the script directly. For example, it's not totally clear what the sleep_VIM dataset is, and what the columns refer to. It would make the computations we perform more meaningful if students knew what the dataset consisted of, what the features were, and what units the columns were in.

Lines 55 and 57 need to be shortened so that they can be tab autocompleted by the users.

Factors are mentioned near the end of the lesson, after the introduction of data frames. This is likely to explain the stringsAsFactors parameter for data frames.

This is likely not necessary. First, stringsAsFactors can simply be omitted here: the default argument is FALSE, so it doesn't need to be discussed at all. It can be brought up in Part 2 if necessary.

The introduction of factors should occur with the other data types. It's a little odd that it's mentioned in the beginning of that section, but not referenced till much later in the lesson.

Line 220, defining the scientific notation, comes somewhat abruptly. It is likely not necessary to demonstrate usage of this function. Its use case is not motivated and it is not referenced again in this lesson, so it makes the momentum of the lesson feel a bit awkward.

.Rproj.user shouldn't be tracked in git as it's a user-specific temporary directory - it would be good to remove.

there are instances where we define a vector X = 1:5, then we do a for loop: "for (X in 1:length(X))". This is super confusing for people since the iterator is the same name (and values!) as the vector. We should use a vector X and an iterator i to avoid confusion. Also ideally the vector X would be X = 2:6 so that the value of X[i] is not the same as i (easier to explain this way)

Workshop title should be "R-Fundamentals:-Parts-1-4"

At some points in the lesson, it appears that conversions from characters to factors are done, or vice versa. These are not done consistently. The default import does not convert the characters in the dataframe to factors, but later code blocks assume that these columns were already factors. They should be made consistent.

It is probably best to import the dataframes with stringsAsFactors = TRUE, and then make sure every conversion thereafter makes sense.

Currently, the final example is rather involved: it requires writing a function for the birthday problem. This is somewhat advanced, and difficult to squeeze into what is already a full lesson.

A simpler example could involve writing a naive function that checks whether a number is prime - this involves a function, a for loop, and an if statement, and simpler mathematical ideas.

Section 8 discusses getwd() and saving a csv file. This comes a little bit out of nowhere, and would fit naturally in Part 2 when importing csv files from a specific path is discussed.

Ideally, this repository would have a lessons folder in which each Part would be placed within. A barrier to this is the importing of data files: right now, they are hard coded according to the path, starting in Part2.R.

A better approach would be to place the files within the lessons folders, and use the here package to properly create the import. This would be a good instruction point, as here is used in basically all other R workshops.

The best place to do this is in Part2.R. However, these changes would require going through all scripts and making the appropriate changes to the filepaths.

Using guides(fill = FALSE) is deprecated, and must be updated to remove the warning.

There is bit of a bottleneck in Part1.R - Installation trouble shooting alone can take upwards of 30min-1hr.

A possible solution is to move some materials from Part1 to Part2 (especially anything dataframe related since it becomes redundant). This can make the focus of Part1 to help with troubleshooting getting started with R rather than getting deep into data structures. Will update this issue after observing Part2 today.

@Averysaurus

It is probably not necessary to introduce random number generation in the very first lesson. It can be confusing for first time users and is not necessary to create dataframes - we can just create the dataframes by hand.

Furthermore, this concept is also brought up in Part 4. It could be moved to that lesson entirely.

This issue is for discussion. Is it worth keeping the statistical testing section after the introduction to visualization? It feels a bit odd because we're not explaining any of the details of the statistical testing - just demonstrating that it can be done. Furthermore, it feels only loosely connected to the previous part of the lesson.

I think it's totally fine to keep geom_smooth in there as a way to demonstrate that linear models can be fitted and plotted in one fell swoop. However, demonstrating the other functions may not be that effective since the lesson does not really go through the weeds.

We define the variable "number" and assign 5 to it but don't use it to demonstrate how to use a variable. Might be good to add one line like "number + 8" to show that we can use the variable like a number (which it is). I would add that right after the line "class(number)"

There is a misdirecting link under Installation instructions. The 3rd bullet "Download these workshop materials:" links to the R Data Visualization workshop. Instead it should link to the R-Fundamentals workshop (https://github.com/dlab-berkeley/R-Fundamentals).

The current Part1.R script errors on lines 354 - 368 because of a typo in the country column (it doesn't capitalize "Country").

Explained named arguments in more detail in Part1-4

In lines 163-195, several functions operating on character objects are demonstrated and used. This is probably not necessary for the very first lesson in R. Furthermore, they don't fit the section, which aims to demonstrate data structures (vectors, lists, and data frames).

They should be removed from this section. They could potentially be excised entirely, or placed in the section which introduces built-in functions.

The solution to Challenge 4 requires using dplyr, which is not discussed thus far in this series. It would be better to use another, more approachable example.

Cylinders challenge was very well done and boosted everyone's intuition a bunch. Similarly the Monte Carlo dice rolls seemed to be useful. The Birthday problem was a great one to end on because it did something that actually demonstrated the power of the technology, rather than giving us intuitive answers that we didn't actually need R for.

The "Lock" example was overkill in its complexity, I did not even attempt it with my group. I suggest to remove it.

Overall, there were perhaps too many examples in this section to be completed in the 3 hour time limit.

I'm reviewing Part1-3 and don't think that we need to mention factor type. As a matter of fact, we didn't discuss it in that particular section.

Useful for checking if a string appears in a vector of strings (maybe part 1 or part 2 with boolean stuff?)

1. Introduce %in% somewhere - it's really useful for checking if a string appears in a vector of strings (maybe in part 1 or part 2?)
2. Part 4 is overall too long and complicated (a few specific suggestions below)
3. Start with if statements, then for loops, then functions (currently it's functions > for loops > if statements > more functions)
4. In the loops section, there are instances where we define a vector X = 1:5, then we do a for loop: "for (X in 1:length(X))". This is super confusing for people since the iterator is the same name (and values!) as the vector. We should use a vector X and an iterator i to avoid confusion. Also ideally the vector X would be X = 2:6 so that the value of X[i] is not the same as i (easier to explain this way)
5. Cylinders challenge is great!
6. The "Lock" example is overkill, suggested to remove (I did not even attempt it)
7. Monte carlo dice rolls & birthday problem were really well received

Thanks!

It was pointed out in Day 4 that 0 is not a negative number. Two of the example if else statements (lines 208-219 in Part4.R) produces results that say it is so ("0 is negative."). Code below. Could change line 216 and 226 to "x is less than 1" for example if we want to address this technicality, or add another else if statement for "x is equal to zero."

Lastly, if we want to check for multiple conditions in a row, we can use else-if statements:

x <- 2
if (x > 3) {
print("x is greater than 3.")
} else if (x > 0) {
print("x is greater than 0.")
} else {
print("x is negative.")
}

Let's go ahead and put this if-else statement inside its own function, and run it inside a for-loop:

if_checker <- function(x) {
if (x > 3) {
print(paste0("x = ", x, ", is greater than 3."))
} else if (x > 0) {
print(paste0("x = ", x, ", is greater than 0."))
} else {
print(paste0("x = ", x, ", is negative."))
}
}

The link to datahub is currently broken:

This is up for discussion. However, I feel that functions like subset and merge, from base R, are not really used by practitioners these days. Instead, students would benefit most greatly from introduction to modern tools based on tidyverse packages such as dplyr. We can mention that these functions exist in base R, but we should instead do subsetting and merging with dplyr functions. This mimics how ggplot is introduced in Part 3.

I elaborated on why we should encourage students to use <- instead of = for variable assignment. For sure, = could be more intuitive as it's the way a variable is assigned in Python and some other languages. However, = and <- work differently in terms of their scopes. I referenced Google and Hadley Wickam's R style guidelines as well as a minimal reproducible example that demonstrates the point.

Line 419:
head(sims)

Needs to be changed to:
head(results)

Part2.R - Line 217
install.packages(tidyr)

Needs to be changed to
install.packages("tidyr")

Was encountering issues with participants and how the DataHub link was formatted, (not loading into R-studio client). Formatted a new link and it seems to work on my end:

https://datahub.berkeley.edu/hub/user-redirect/git-pull?repo=https%3A%2F%2Fgithub.com%2Fdlab-berkeley%2FR-Fundamentals&urlpath=rstudio%2F&branch=main

I think that the here package is worth mentioning even for the beginning users as getwd() + setwd() is not a good practice for reproducibility.