My 2 cents about Lab1: Q5

[Darth Knight]

Because of the outliers, the boxplots are squeezed
and become incomparable each other.
This can be resolved by adding "outline=FALSE" argument into "boxplot" command.
It will remove the outliers so that you can easily compare the two boxplots.
Let me know if this works okay.

[brice]

This helps very much, but I don't understand exactly what the addition tells it to do and how it affects it

[Darth Knight]

It is just an appointed argument to decide whether
to see the outliers on the boxplot or not.
The default was "TRUE" which means showing; so, we can make it "FALSE".

In general, "?command" will show us its help document.
For example, by using "?boxplot", we can find the explanation
about the argument as follows:

"outline if outline is not true, the outliers are not drawn"

In many cases, the help documents can save our googling time a lot!

[MegDawg97]

For histograms the outline=false command doesnt get rid of the outliers, how would you do that for the hist?

[Darth Knight]

Yes, there is no such luxury in the default histogram command.

If we really want to get rid of the outliers in histograms,
we need to either find out a special package which would give you an option for the outliers
or remove the outliers in our data first and then plot it using the default histogram.

Since I cannot recall a specific package, let me introduce a function which removes the outliers in your data:

remove_outliers <- function(x, na.rm = TRUE, ...) {
  qnt <- quantile(x, probs=c(.25, .75), na.rm = na.rm, ...)
  H <- 1.5 * IQR(x, na.rm = na.rm)
  y <- x
  y[x < (qnt[1] - H)] <- NA
  y[x > (qnt[2] + H)] <- NA
  y
}

If you run this entire code in RStudio and use it by "out<-remove_outlier(YOUR DATA)",
then "hist(out)" will give you the desired histogram with no outliers.

Lastly, the outlier detection rule used in this code is called "1.5*IQR rule".
You can find an explanation here: https://www.youtube.com/watch?v=FRlTh5HQORA