Purpose: This notebook will help you remind yourself of the following items.
{tidyverse}.{tidyverse}?Since you already have exposure to working with data in R, you’ll have a pretty unstructured analysis to do. That means you get to decide what interesting questions to ask and how to answer them. Choose either of the following data sets to utilize.
The hits and home runs data perhaps provides a greater challenge than the FAA data set since the MLB data is split across three tables. Choosing the MLB data set will give you the opportunity to practice with joins.
Choose which data you’d prefer to work with. For convenience, I’ve extracted smaller versions of these data sets and uploaded them to one of my GitHub repositories here.
Use File -> New File -> R Markdown... to create a
new R Markdown notebook. Clear out the boilerplate and use the
setup chunk to load the {tidyverse} and
{tidymodels} libraries. You may also want to load
{kableExtra} for printing out nicely formatted tables in
your knitted notebook.
Loading the data from a web URL is convenient because you don’t need to download anything. You’ll need the following files depending on which topic you’d like to explore:
faa_birdstrikes.csv)battedballs.csv and
park_dimensions.csv)To load the data, you’ll click on the file name and then look for the
button labeled Raw in the grey banner above the displayed
data. Click that button and then copy the URL from your web browser (the
URL should start with raw.githubusercontent.com/...). In
the setup code chunk, use read_csv() with the
URL in quotations to read the data into your notebook. Don’t forget to
use the arrow operator (<-) to store your data into a
named object, and don’t forget to read in both files if you are working
with the baseball data.
Create a new code chunk and use it to print out the
head() of the data set(s) and provide a description of the
overall context of the data.
You’ve had previous exposure to statistical modeling and you know of the importance in splitting your data into training and test sets. We use test observations to assess our model’s performance in an unbiased manner. These assessments are only unbiased if we don’t know anything about the observations in our test set though. We need to be very careful that neither we, nor our model(s), learn anything about the observations in our test set.
That being said, there are still some questions worth asking about the full data set. For example,
Why might we want to know answers to these questions? Answer them for your data set.
Once you have these answers, use initial_split() and its
corresponding helper functions to split your data into training
and test sets. Look into the strata argument and
determine if you should be using it. Use it if so, and don’t forget to
set a seed so that you are always obtaining the same training
and test data.
Univariate questions are questions involving a single variable. These types of questions are useful towards gaining insight into the distribution of your variables. In particular, understanding the distribution of the response variable is important (as it was above). These questions can also give us insight into whether the random training set we’ve obtained is somehow not representative of the overall data set we began with. For example, are some categories vastly over- or under-represented in our training set?
Justify asking several univariate questions about your training data and answer them. Be sure to include narrative text and not just code as you run these analyses.
Multivariate questions are questions involving multiple variables. These questions are useful in finding connections/associations between variables. In particular, we might be interested in which variables are associated with our response variable.
Justify asking several multivariate questions about your training data and answer them. Be sure to include narrative text and not just code as you run these analyses.
Once you are done, use the Pull -> Commit -> Push
workflow from the Git tab of your top-right pane in RStudio
to send your updated file to your remote repository.
You’ll return to this notebook next class meeting.