Getting Started with R - Part 4: Vector Extracting, Replacing and Excluding

Reaching beyond the basic indexing of Vectors we started in our previous post Vectors Operations we will continue selecting, subsetting, replacing vectors beyond simple indexing.

I am posting this tutorial as I learn R. I will respond to feedback for errata in the comments.

Extract with the Extract\Replace Operator []

Some recap from earlier lessons

Extract via index

In Part 2 of the tutorial we saw that we can use the [] operator similar to indexer in tradtional languages by providing a 1-based index. Like this:

combined_integer <- c(1, 2, 4)
combined_integer[3]

But as we saw in Part 3, we need to think beyond our old ideas of indexing. We can just as well pass a vector to the [] operator. For instance…

combined_integer[c(1,3)]

…will return the first and third values.

Often we want to select contiguous values. There is a quick constructor to help with that in the : operator. Try this

1:10

That was essentially the same as calling c(1,2,3,4,5,6,7,8,9,10)

cincinnati_rainfall <- c(2.87, 2.64, 3.82, 3.82, 4.72, 4.17, 3.86, 3.98, 3.11, 2.83, 3.31, 3.11)
names(cincinnati_rainfall) <- c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec")
cincinnati_rainfall[3:8]

Gives us a nice concise way to get our rainfall from March to August. Surely more convenient than typing cincinnati_rainfall[c(3,4,5,6,7,8)]

Extracting via logical

In Part 3 we also learned that we can pass in a boolean vector to select. Just a recap:

We created two vectors

weekday_names = c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday")
predicted_precipitation <- c(0.0, 0.3, 0.01, 0.07, 0.0)
names(predicted_precipitation) = weekday_names
actual_precipitation <-  c(0.0, 0.27, 0.0, 0.08, 0.02)
names(actual_precipitation) = weekday_names

which we could then subset (extract) like this.

more_than_predicted <- predicted_precipitation > actual_precipitation 
predicted_precipitation[more_than_predicted]

The logical values in the more_than_predicted vector acted as slector that was applied in a one-to-one fashion and where the selector value was TRUE the value at the applied index was returned

Extracting by name

We game our indices names, so can we use them with the extract operator? Of course we can

actual_precipitation["Wednesday"]
predicted_precipitation["Friday"]

This is certainly very nice and readable.

Self Study: Use a vector of characters to select Wednesday and Friday’s actual precipitation.

Replace with the Extract\Replace Operator []

Just as we can yield values or a subset of our vector we can set them with the same operator. Lets update our prediction for rain on Thursday and Friday to 0.0

updated_predicted_precipitation <- predicted_precipitation
updated_predicted_precipitation[c("Thursday", "Friday")] <- 0.0
updated_predicted_precipitation

Two for the price of one! What if I want to set all my values? Our updated prediction is for no rain can I just set them all like this?

updated_predicted_precipitation <- predicted_precipitation
updated_predicted_precipitation <- 0.0
updated_predicted_precipitation

Nope. Lets try again. This time change one thing, we add []

updated_predicted_precipitation <- predicted_precipitation
updated_predicted_precipitation[] <- 0.0
updated_predicted_precipitation

One simple way to think about replacing is that whatever was we received with the subset yield, is replaced with the right hand side of the assignment operator. Its easy to make mistakes. Let see we want to pad our predition for non-zero days.

padded_predicted_precipitation <- predicted_precipitation
padded_predicted_precipitation[ padded_predicted_precipitation > 0 ] <- padded_predicted_precipitation + 0.5

Did you spot the mistake? The left hand of the assignment is a smaller vector than the one on the right. padded_predicted_precipitation+1 has 5 elements and padded_predicted_precipitation[ padded_predicted_precipitation > 0 ] has 3. Lets fix our code

rainy_days <- predicted_precipitation > 0
padded_predicted_precipitation <- predicted_precipitation
padded_predicted_precipitation[rainy_days] <- predicted_precipitation[rainy_days] + 0.05
padded_predicted_precipitation

Thats better. There are no compound assignment operators in R as far as I know… so no += equivalent.

Excluding

Lets look our rainfall vector again

cincinnati_rainfall <- c(2.87, 2.64, 3.82, 3.82, 4.72, 4.17, 3.86, 3.98, 3.11, 2.83, 3.31, 3.11)
names(cincinnati_rainfall) <- c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec")

What would you expect for cincinnati_rainfall[-1], Try it in the console and see what is returned. If you come from python it may be a bit of a surprise. - in the context of our extractor operator means exclude, but we can’t use it with labels or with our range generator, so these don’t work:

cincinnati_rainfall[-"Jan"]
cincinnati_rainfall[-7:12]

but, we can exclude index vectors using the c() function. Which fixes the issue with the range generator.

cincinnati_rainfall[-c(7:12)]
cincinnati_rainfall[-c(1,3,6)]

As you can see we exclude the last 6 months in the first case, and January, March and June in the second case.