An Introduction to Data Visualization with RStudio November 29, 2018

Transcription

1 Good afternoon everyone my name is Bailey Maryfield and I am one of JRSA's research analysts. For those of you less familiar with JRSA, that stands for the Justice Research and Statistics Association. We're a national nonprofit organization dedicated to the use of research and analysis to inform criminal and juvenile justice decision making, and we are comprised of a network of researchers and practitioners, which at the core include the directors and staff from state statistical analysis centers. Before we go any further, I want to thank our partners at the Bureau of Justice Statistics for helping to make this webinar possible. With that, it is my pleasure for me to welcome our webinar today and this will be on an introduction to data visualization with RStudio presented by Dr. Stephanie Santorico from the University of Colorado Denver. Stephanie is a professor and director of statistical programs in the Department of Mathematical and Statistical Sciences and is a statistical geneticist who has worked for over 20 years on the development of statistical methods for the use of inheritance and humans to understand human disease and health. She also serves on the genomics computational biology and technology study section for the National Institutes for Health. Stephanie has a wealth of experience in using and teaching R, so we are very happy to have her here today to lead the webinar. As you can see, Stephanie will be using her video today to make this webinar as interactive as possible and again if you have questions throughout, please just use the chat box or raise your hand and I can unmute you. With that, Stephanie welcome and I'm going to now turn it over to you. All right, great. Thank you, thanks for the introduction Bailey. Can you guys hear me okay? Is that all right? Am I on? I'm assuming I am on. Yes, you sound good. Oh good, okay. All right. The focus of today as courses as it shows, R in RStudio and in terms of thinking about this and talking with Erin Barley to invited me, we decided to start with the assumption that you are not an R user necessarily and so what I really want to do is to get you in, get you started on R, and more specifically in the context of RStudio. That's why if you have questions as we go along, please do feel free to ask so we can make sure you're able to move along okay. I have slides for you to use. You can download them so you'll see the first link. It's for these slides. If you would like and I encourage you to follow along as I do the codings, I'm going to show you how to load the script, how to load data. There are another two links down there and so I would encourage you to pull those down and make sure you know where they go on your computer on so that we have access to them. With that, let's get started. This was the abstract you guys would have seen on the announcement, but the first thing we're going to start with is just get you introduced to really what's the value of R and more Justice Research and Statistics Association Webinar Page 1 of 15

2 specifically RStudio, what they are, why they're worth knowing about as well as we'll give a few comparisons to other tools that you might be using or you might be thinking about using. I'm going to get you to where you can read in that dataset, so this will start with the dataset on marijuana arrest. I will start by saying I do not do criminal justice research. You can forgive me if I make any ridiculous plots, but the point was that Erin thought at least this was some data that you might be more familiar with and we could use some of the variables to do plots. They may or not make the most sense of the set of data, but at least it gives you a point of illustration. Hopefully by the end, you're going to be able to take the script I provided and actually recreate some of these plots among your own data, so that's the big goal. If we can get you there on some range of simple plots that we'll start with, that will allow you to be a little more comfortable when you start moving into a lot more advanced plot options and there really is a ton that you can do with R. I'll mention a few of those at the end just to give you an idea of the array of things you could go to from here, so that's the purpose for today. First off, I don't know how many of you have used R. Again, we were assuming a minimal amount. R is a free language, so the free statistical language that allows you to do methods, to do graphics, to do computing. One of the perks is of course that it's free, my students love it for that reason, but it also "easily" runs on a lot of different platforms. I mean it's been a game changer in terms of statistical research. When I started 20 years ago, we really didn't have such access to so many different methods and functions. It's become easy to make your own functions. You can make your own packages and the user base is incredibly large. If you have someone who is creating new methods or new plots, they can easily turn around and make those available to others to use. Whereas in the past, it can have been a difficult process of making executables, making them available to different platforms. R really changed how accessible advanced methods and graphic techniques are, so that's a huge benefit. Now R is the language and depending on how comfortable you are with programming language, you could just go straight into a console and start typing away. Most of us like a little bit of a better interface and that's where RStudio comes in. R is the language. RStudio is a development environment. The real perks of this at least from the user perspective, it gives a nice Windows interface. It gives access to some point and click themes. It has tools for plotting, for history. If you are a person who likes to get into a more of a debugging sophisticated interface that has that option and have some workspace management. RStudio gradually I think even the programmers eventually move into it because it has coding perks as well including syntax highlighting, direct code execution as Justice Research and Statistics Association Webinar Page 2 of 15

3 well as being able to edit a script as you're moving along. For us, for the purpose of the webinar, I provide this quote from the R project website which I think is spot-on. One of the biggest strengths for R is also being able to produce welldesigned publication quality plots. When I'm publishing articles, R typically is what I used to be able to get a high enough quality plot and to be able to really manipulate it to represent what I'm trying to stay within that plot. Huge perks for being able to use that. Before we get started, I want to give you an overview of what RStudio looks like. If you have downloaded it, we're going to move over and actually take a peek, but you've got generally four quadrants. Upper left is where we're going to have scripts. Scripts are fantastic in terms of if you are using a script, you're going to be able reproduce the analysis that you did so that is a huge benefit. I saved increasingly recognized in the field to be able to reproduce your results. You would be able to hand that script over to someone else and they would be able to know exactly all of the different alterations you made to your dataset into your analysis, so that is a huge plus. Bottom left, you'll see you've got a console. This is a place where you can just directly type in commands. On the top right, you got access to the environment. I'm going to click over in just second for you to look at and in the bottom, you get access to thoughts, to help actually to your file structure. There's a lot you can really get into just within the RStudio package. Just to illustrate that, let me switch over so you can see. All right. Hopefully, you're now seeing the RStudio screen. If you did download my script, right now you unless you went ahead and opened it, you're not going to have that here. What I would encourage you to do so that you can get the script up, go to the menu where you see file, click that, and open file and then navigate to where you downloaded that script to you. I had already downloaded this, but for me it would be in this directory. I would just click on it and then open it. It would take a minute to open that up and that will show right here and so it's a script. What that means is it's going to be a set of commands for you to go through and execute. Nothing has been done yet. When you open it, it's just like opening a text file which also means as we're going along, if we decide to make changes, we can make changes within that script, save it, and have access to it to use again. A couple of things you'll notice. I tried to make the script overly annotated, so that you can really see what each of the things is doing. I'm going to step through the script as we go along. You'll notice these little hash marks, so anything beyond the hash mark is treated as a comment. It's ignored in code and so that grayed out area is going to be where you're able to see my instruction on what this is really doing for you. Bottom, so this is where we can type theme. We can do a simple map, so two plus two. You can have access to functions. We are going to actually execute what's on the top and it's going to run within this bottom console window. Other things, so the upper right, this is data so I've already read the Justice Research and Statistics Association Webinar Page 3 of 15

4 data end and one of the nice things is that in RStudio, I can now actually look at that dataset. I opened up, I call it that like almost everything that because I apparently lack creativity, but it also makes my coding a little reusable if I keep it generic. I just opened up this marijuana dataset that we're going to use and you can actually go through and look at it within RStudio. You can get to a history of commands. If you forget what you did and you just want to go through and remind yourself, you can get to your history. Right now I don't have any clocks open, but if I did like I might create one really quickly, those are going to show here. You can export them, so you can export them to PDF to different image types. You can copy them. If I'm creating plots and I'm going to it, I might just copy it to the clipboard and send it to someone to have a look at, so that's a nice ease of use for getting your plot. Packages help, so there's a lot of access to information just within this environment, so that's one of the big perks for it. Let's go back to the presentation and the slides. That's what RStudio looks like, so we're going to be using that. Before we jump in, you might want a little bit of motivation behind how it compares to other tools that you will be thinking about and I say right now the big tools in terms of data visualization are R and Tableau, and then maybe a little bit older school but a lot of utility can be found for some of the basics just in Excel, and so depending on what you're working with. I'm going to give you a couple of comparisons. In terms of Excel, so Excel is nice and that it gives you point and click and it does some easy statistics for tabular data. If you have row those observations and columns as variables, Excel can easily work with that. R does the same functionality that you would get out of Excel, but certainly gets much more into sophisticated analyses. R does not automatically have point-and-click, whereas Excel does so it's going to make it a little bit easier to use. Frequently unless you're just a programmer at heart, R is going to have a little bit of a learning curve, but certainly sticking with it, you're going to have access to a lot more sophisticated methods and graphs. Hopefully we're going to get you at least partially a half for that learning curve. One of the frustrating points I would say frequently with Excel is there's not an easy history of all the things you're doing to a set of data and again, I think reproducibility is increased in a CNS being very important. One of the benefits for R is you can start from a raw dataset and have a script that describes everything you did to that set of data, all your filtering, any changes you've made, all of the analyses that you've done which would include the assumptions of both analyses. While Excel's point-and-click is easy, I'd say the R script access is a huge plus. Tableau, so if you are not a programmer and you want some quick fairly sophisticated visualization, Tableau is a great option and you don't need to be a coder. It's point and click. R does allow for many if not all of the same plus you Justice Research and Statistics Association Webinar Page 4 of 15

5 would get out of Tableau, but I'd say a larger array and you're able to do a much more manipulation within R of your plots. If you are wanting to have points look a certain way, if you were wanting to layer text, if you are wanting a 3- dimensional plot, you want it in a specific angle, these are the things you could do in R that you might not have access to in Tableau. Tableau allows for some statistical analyses if they are a lot more, especially sense you have access to the packages and libraries that come along with it. Tableau allows for dashboards to be built and on top of databases, so that is a huge perk especially if you do not have a coder or you're not a coder yourself. R actually allows for that as well. It may be a little more of a learning curve to get into it, but there's a package that allows you to do dashboards also. If you are just starting out and you're thinking about I really want to get more into data visualization, I would say my advice is try out Tableau. It's not free so that's going to be possibly a downside. I'm going to try out Tableau to get quick access to some things, but then start using R so that you can learn to the point where you can create more customized on data visualization. I think both tools can be used. This is just a quick comparison between a couple of options. Let's go ahead and get started, so the very first thing we're going to do is read in our dataset. We can read in a variety of formats within R. You can do this both through the drop-down menu and I'm going to show you how to do that for today. I'd say again reproducibility wise, if you are using a script, it's good to actually read your data in there because you know where you read the data. If you have a project structure, it's going to show you exactly what directory within that project structure on your computer you pulled that data from. Having been on many projects where there have been multiple iterations of data as new data comes in, that at times can be a lifesaver to know where did those results come from, which particular version of the dataset. That's a big positive. Let me share. If you want to go ahead and switch over to R. All right. Go into the file drop down menu. There is an import dataset and so the dataset that I gave you will link to... I'm going to give a little more of a description as to what that data is, but it's a text file. You're going to want to pick some text that first option and then within the window, navigate to where you downloaded that data. For me, it was in here. I'm going to click on the specific dataset and open that. You should see a window that gives you some options to specify the format of the dataset and one of them has the name... Like I said, I usually call everything that, so I'm going to put that up there. You can see the file is trying to input on the upper right. It is comma delimited, so this is an option. If you had an Excel spreadsheet and you wanted to save it, you could save it as a comma delimited file. All of the things in here work for us. Comments a separator, yes. Periods are used for decimals. Quoting protects. Justice Research and Statistics Association Webinar Page 5 of 15

6 One thing you always have to make sure you understand this for your data that how we're missing values coded and so for this dataset, they used capital U and K for unknown, and then they also used a capital U, so depending on the variable. If you can see that what I did was I in the streams box, I typed capital U and K comma and then capital U, and so that will treat any of that data as missing as a thread in. Then you just type import and it will bring your dataset up for you. Actually if you look at the bottom left, it tells you the command it executed. This is a sneaky way to learn the coding because the point and click menu executes and shows you the command down at the bottom, so that will get your data in. Let me take a pause and Alyssa, "Can you please restate how to download the script and the data?" Absolutely. Go to the first slide we've started on, so that's slide five if you're moving back and forth. These links at the bottom are how you get to the script in the data. This is for Jamie, so tinyurl.com and then R tutorial script one and R tutorial data one will get those downloaded for you. All right, so let me go back. Anyone else having problems opening RStudio, getting the data into R? Looking at the chat or perhaps. Okay, I'm going to assume we're okay, but please don't hesitate to let me know. I'd hate for us to be moving along and you're not able to follow, so let us know. A few basics if we start digging in. Like I said, we're going to use the script so you will be able to open that script. When I go back over, we'll make sure we're all on the same page there. We are going to be making use of some packages so this is going to give you an example of number one, how you would install the package and number two, after you've installed it just one time, you have to load it into R anytime you're using R. It's not going to have it loaded unless you need it. We're going to look at our dataset, but an overview. When you read in a set of data, R treats it we call it a quality data frame or data matrix and one of the nice things is that you can get to so if I wanted to look in that dataset and I type something like that live IJ, that would give you the element in the I row and the J column. I can set out an entire row by just putting in a row number. I can pull down an entire column, so that typically would be a variable by just saying comma J. You'll notice these are square brackets, so this has access to the data that you've loaded as a matrix. Another thing you're going to see me doing is accessing the variables. Within that dataset, each of my columns has a different variable and they have names. This dollar sign symbol is another way to just say hey, I want to get to the age variable within the dataset that, so we're going to use that structure. I'm going to use a number of functions, so there are functions that are built-in. They're going to be functions that we are able to use from the package as we load and just to give you an idea of what those look like, so it's going to be the function name, so whatever it is and then parentheses and then you'll have options for the function within those parentheses. We're going to use a number Justice Research and Statistics Association Webinar Page 6 of 15

7 of functions as we go through. Each of our plots are going to have a function that get that specific plot, so that's what we'll be working through. This is the dataset that I'm using, so this was I recommended by Erin. Again, I don't search marijuana, I'm not a justice or criminal researcher, so bear with me and forgive me for anything ridiculous that we do. Purpose is just to give you guys some plots on data that you might be interested in playing with. The website here gives you access to where the data was pulled from which has more information about on what was collected just to give you like the first part of the PDF that describes the dataset. These are from the District of Columbia and so each row is a different arrest and they've got variables like an X, Y location for where the arrest was made, the quantitative variable that they had in there with age. They have sex of the offender. They have the type of arrest that was made and they have some other things in there that you might want to look at, so that's the data we're going to use today. The file that I directed you to for was the adult data. I've screened at the juveniles because a lot of the data there for privacy reasons have been unset and missing, so we're just going to look at the adults. All right. Plot, so we're going to start with some I would say "simple," I would say more standard plots. Histograms, box plots, density plots, 2D scatter plots. We'll get a little more exciting from there, so I'll give you some options to do a 3- dimensional plot that is static, a 2-dimensional plot that's interactive so you're able to twist it and look at it. That has been very helpful at least for me for exploring the patterns I see in my data. I'll also show you a more advanced command you can execute that will save that to an HTML, so you could come back to it or share it with someone else and they are able to go in and manipulate that data. Then to get you a little bit closer to something more sophisticated, we're going to do hexagon plot and thanks Bailey, got that. All right. The way this is going to work is I want to go back to the script to show these things interactively. The slides are really more of a reference for you because if especially if this is all brand new to you, it may look well like a foreign language, but you should be able to come back to these plots and see on the left, I'm giving you the plot and on the right, I'm giving you the specific part of the code that generated that plot. I'm going to show you the default options which are not particularly sexy and then we're going to put in some graphical parameters that will make that look a little bit better to use. Let's go back over to R. Hopefully if you're wanting to interact and go through this, so two things, make sure you have loaded the script. You can push this little folder button this will third one over to open and then you can go to the directory where you downloaded the script. Here's my script that I actually already have it open, but there we go, so that would open the script that I'm using. Data will make sure you read the dataset end, so the script for me I have the command that reads it in, but notice that it's specific to my computer because it tells what directory it's in. Justice Research and Statistics Association Webinar Page 7 of 15

8 What you might want to do is just read it in through the menu, so file, import data from text. Again, go to where your dataset is and then the really important thing is there, all right so locate the data so that you have the unknowns. Leave everything in here the same except for the name. Change that to that so it will work within my script. In NA streams, make sure you tie it capital U and K comma and then capital U, and that should read it in as that, right. Once you have that up, go over into the script and take your cursor down to that line that says hist.. Yeah, a dot dollar sign age and you can select that. Wait, when you have your cursor on the line, if I go up to this run button on the top right, it will execute that line for me. Let me make sure I've got my data in there right. There we go. Give that a try, so put your cursor on hist, push run and that will run the command, and then you should see the plot pop up on the bottom right. Bailey I can't see the chat window so if anyone has questions, will you please make sure and interrupt me as we go through? Yes, will do. Great. Okay, so that's not particularly pretty as far as a histogram to show people. You can change basically everything about that plot and so in my command or in my comments right there, you'll see I'm going to change the X label to say something better dot dollar sign age. I'm going to change the top of the plot for that hell of it I'm going to change the color or between marijuana so I picked start green, you go with that. If you want to see more options, I put a nice link into the comments that will show you the different lines, the different, points the different colors and there really is a lot of options for the plot. Click your cursor if you're going along on the hist line and then again, I'm going to hit that run button on the top, and that will create a much nicer looking plot. I changed the Y-axis count to X-axis to age. I put a better title on it and then a nicer color. If you want to compare them in the plot thing on the bottom right, there's a little arrow you can navigate back through your plots. Maybe you forgot to save or copy one. You can navigate to any of the plots you've created, so that would give you a basic histogram. If you wanted to do a box plot, okay so notice so the function is box plot. It's a function because you can see my parentheses here and then I'm doing a box plot of age and I know that because it says dot dollar sign age, so that was the one quantitative variable in the dataset which is why I'm using it. I hit run and that will give you a box plot, right? Nothing particularly exciting because it's not even labels but if you wanted to change the color, you could use the same COL. You could label the Y-axis, label the X-axis. The nice thing about R is the options transfer in between functions, so you can make that a little prettier. If you wanted to do a group to box plot that has a nice option. Justice Research and Statistics Association Webinar Page 8 of 15

9 First off, this is a table command so I just wanted to see how many men and women there were. The dot dollar sign and then defendant sex pulls that sex column from my dataset and table just creates a contingency table for it. You can see if you look on the bottom left, there are a 1090 female adults within the dataset and 8807 adult males. In the box plot, you'll notice the tilde here. The box plot if you put the variable you want represented in the box spot, the tilde and then a group variable, it will do side by side group the box plots, and even though a simple plot frequently it's very helpful to really look at the distribution over groups. If you put your cursor on box plot, you type run, you'll see I get side by side, they're actually labeled. Like I said, if you wanted to go in, do the color, you do a title, you could do all of the same options to make that a little more interesting. Those are some box plots. Also Erin mentioned that you might be interested in density plots and so in here, you'll see I've got actually a nested function that I'm running. Inside you'll see and I'm highlighting it, so density. What this is going to do is it's got reasonable default parameters, but it's going to create a density for the age variable. I'm removing the missing values, so that that doesn't trip the function up and then on the outside, you'll see plot. What this is going to do is take your data, create a density function for it, and then it's going to plot that function. If I put my cursor on that and I hit run, it comes up. Again, I just use default option just to see how it works. We can make this fancier with any of the same options I show it when I did on the histogram, right. Maybe more interesting actually would be to do a density plot where you separate based on groups, so we could also do that. Look what I have code for you next and one of the themes that I talked about how to access data within the data frame. You can actually create boolean vectors to subsets. If I were to put a boolean for a true/false variable within the row designation, what that would do would be just to pull those values that have a true value for that boolean function. From coding here, I'm highlighting. What this is doing is it's saying okay inside these parentheses, if the defendant sex for my dataset is male, then it's going to define a variable men which is called it's true if that's true and it's false if it's not. Same thing for women. If the defendants sex is female, it creates that boolean true for female, false for males. If you either highlight that or just execute each line, that will define those variables. If you wanted to see what it looks like in the console window in the bottom left by type men, right. It'll only spit out so much, so it protects you from trying to throw out 10,000 variables to your screen, but you'll see I've got true/false options for each individual, whether they are a male or not. Then I'm doing the same density plot only in here, so I'm highlighting. You'll see that instead of just having dot dollar sign age in bracket, I have men comma and so that's going to subset my data to just men and just plot the density for their values. I added on a color of red. If I run that, so you'll see I now have a density plot. If I want to add another plot on Justice Research and Statistics Association Webinar Page 9 of 15

10 to that, I use the lines commands and so you'll see I'm doing line and now I'm going to add on the dot for women for age. I colored it blue just so we could see the difference between those. It's texted on there, so the distribution here is not that different. Again, we just pulled the dataset to use for graphing purposes, but it doesn't look like the age distribution is too different for the male and female offenders. We can make this fancier, so I'm giving you a few other options, so I change the color. You can change the line type and so that link I had in the comments will show you all the different lines that you have access to. I can change at least for my eyes, those plots are a little bit skinny so you can make them bigger. I'm redoing the plot with some of those options. All right, so it's a bit better. I change the X-axis label. I put a main on there and then you can have a legend which of course you should. With the legend, once I have my plot, I will look over here and find a nice white space and you find the upper left corner. Here I said oh well when age is about 60 and the density value is around 0.05, that would be a reasonable place to put a legend box. This is something you have to play with till it looks visually appealing, and then I'm telling it so my legend I plotted men and women. Their colors differ by red and blue. Their line type, so I did a one and two for a line type and the line width was two, and so that will create them a legend for you on your plot, right. A few things for you to add on, but that will get you some density plots. A legend function is generic, so you could use that on other plots as well. Even the box plot, you could add on that to the histogram on, so you've got some nice control within your plots using a legend. The next one I wanted to show you is a scatter plot and so for this dataset, there were X, Y coordinates given. What I did, so you'll notice in the plot, you just put your X variable first. T that long name is what it was called and then you put your Y variable next. I'm going to so again select on that hit run and you'll see so clearly there's a boundary in terms of where arrest location locations are, but it's showing you then just for each point is a different arrest location within this area. You can get stamps here, so the next thing I show you is a few other options. I think I just change the X and the Y labels with a little better. I change these of PCH, changes the plotting character. I happen to like a subtle around circle, but that link gives you access to diamonds and triangles filled in and empty, and all sorts of other plotting characters, and so you can change that right here. Even fancier, what you might want to do is color code these points. One of the variables that they gave us the arrest type, and so you could imagine maybe I want to create an X, Y plot, but I want to color code according to the arrests type. You could get even fancier by changing the size of the point if you wanted to add a fourth variable. Other things you could be doing here. It looks ugly, but I'm going to step you through it. What I'm doing is I'm defining something called natype.col and that Justice Research and Statistics Association Webinar Page 10 of 15

11 is going to be a vector where the color is changing according to the arrest type. I start so this REP is just saying repeat the phrase black, so all of the points by default are going to be black. The dataset had 9918 in them, so it's just going to create one vector of 9918 black, black, black, black, black all the way down. Then I layer on top of that and I say great, so there's another use of boolean inside these square brackets. What this is saying is if in my dataset the marijuana arrest type was distribution, change the color to blue, so that's going to update that vector. If it was possession, make it red. If it was possession with intent to distribute, we're going to call it green. What that's going to do, I'm going to go through and run each of those lines. It's creating that vector that is going to be just colors which change according to the arrest type, and what that means is when I go into plot the location, if I say so COL and I use that variable, it's going to plot each point according to the color I just defined which was based on the arrest type, right. If I run this plot line, run, there we go. Legend, let's add a legend so I broke it up a little bit, but it's very similar. Let's add the legend. Again it's a real dataset, so trying to look at it for patterns, it does look like that public consumption. There is some clustering, so I can see some blocks of black. Whereas in other locations, you're not seeing public consumption arrests, so it's a nice way to add a layer of a third variable on there. Okay, so that will get you to colored basic scatter plots. Let's up it a notch, so let's do a 3D plot and if this is your first time using, so I'm going to enter in here. This is a package that someone has developed, so you need to install it. You can do that with this line, so install that packages and then in quotes, this is the package name. It's a scatter plot 3D. If you execute that, that is going to go and download that package into your storage and that you only have to do that one time, so that just pulls down the coding that you need. The library command that I have my cursor on, that is actually going to load it into your R session so you're able to use it. You need to do that every single time you wish to use that package, so I'm going to run that and what this does, so this is nice and then it will generate a static 3D plot. What I pulled out for here is I'm using the X variable location, the Y location, and then I added age with the other quantitative variable in the set. I labeled my axes similar to up above and I actually am using that same color coding in terms of arrest type, so you're actually getting four different variables within that plot. If I run that that, so it's not rotatable but you can see maybe I didn't pull out a lot of patterns here. In your dataset, I would hope you've got a reasonable three variables that you could look at a relationship with. For me, this would be fine if I was giving a presentation or if I wanted to just show someone how the three variables were related. If I'm actually doing an exploration, my preference is to use a different package to generate 3-dimensional scatter plots and that package requires Justice Research and Statistics Association Webinar Page 11 of 15

12 what's called RGL. Again for the first time first time ever, you will want to install it, so that's what this line is doing for you. If I run that, it's going to pull that package down and install it onto my computer and you only have to do that once ever. The library can be next. When I execute that, that means for this particular R session, I'm going to use this package and the function there is called plot 3D, so let's get a scatter plot 3D which was static. Plot 3D will give you the option of being able to rotate on that and I'm plotting the same thing, so X location, Y location. If you move over, you'll see age is in there, so those are my three variables and I use the same labels. When you do this, it's going to open up a window for you, right. Bailey are you guys able to see that or do I have to share that other window to see the plot? You would actually have to share that other window. All right. Let me share that so you can see it. Question, oh there we go. Can you see it now? Yes, we can. Good, okay. I'm just using my mouse, so what I'm able to do is start to spend that. Again, for me this is helpful in terms of trying to recognize patterns or outliers that I have in my data. I'm going to actually color code this as well. Options, you've got three variables. You can color code out a fourth variable. R has the capability of changing the point size, so you can add a fifth variable on there as well. It really gives you access to some nice exploratory data analysis. Let me do color version. All right. The second plot 3D statement that I have in there, we'll do the same exact thing but color code your points. We do that. All right. Then you can start to try to see if is there any difference in location for X and Y, differences in the type of arrests, differences in the distribution of the age. You start to get some nice options there, so that would get you to a 3D plot. Things that you can do, let me go back to... What I have done in the past is I have rotated it to the point where you can really see the pattern, then I save that as an image to be able to use in a presentation. This is a little extra fancier, these are our markdown package and you execute these two commands that will actually save that into an HTML file and you can send it to someone and they can rotate it and look at it, so that's a pretty nice utility. The last plot I was going to go through is to 2D hexagon plot and for that, you need the ggplot2 library, and this is a brief step into ggplot2 If you were to look for further options for data visualization, ggplot2 is going to be one of the primary ones I would recommend. I've already installed it. You see in the comments, you need to install these packages and ggplot2 is a bit different in Justice Research and Statistics Association Webinar Page 12 of 15

13 terms of syntax. I am for the sake of time just going to show you what it looks like with the commands and let me pop this out a little bit more. You've got the X and Y location, the function itself goes through and defines both hexagons, and then the shade, so you'll see a legend on the right, shows you the density of the arrests in those location with some spots where we actually didn't have any arrest on recorded in those hexagons. That's a nice step into more of an advanced plot. Let me go back over. As I said, I've got your code and your plots within these slides. You're going back and trying to remember which part of the codes and which this should help you out and it steps all the way through the 3D plots, the interactive, and then your hexagon plots. There are many more options you can get into. If you like violin plots, ggplot2, actually all of these heat maps, the geospatial and thematic plots. If you're looking for hierarchical clustering trees or a heat map with the clustering trees, all of these options are available within R. I tell all of my students Google is your friend here, so if you are like I need to produce this specific kind of plot, generally just googling it will give you a ton of resources that show you exactly how to do it. Once you're going to get into it and you start getting a sense of how this works, you can quickly find some good code to reproduce plots to use. I often find quick R to be a good resource and I give you the address for that. They give you some tutorial for how to do different kinds of plots and the basics of plotting, so a lot of it with the things that we've gone through, you're the expert on your data. These spots will go a long way in demonstrating what you're trying to stay with your data. With even more time spent on learning R, there's so much flexibility. You really can tweak things to make the point you're trying to make. If you are interested in getting into, so practicing through these, you want to get more sophisticated, the next step up with the ggplot2 that has so many amazing data visualization options. It's a little bit of a different take on how to plot things, but I strongly encourage you to have a look there. If you're interested in creating an interactive web apps, dashboards, Shiny is a package in our that allows you to do that and then there is a whole collection of different packages more specific to data science through something called the Tidyverse and so I'm giving you the link there as well. Lots more for exploration. Hopefully you're going to be able to take these and reproduce them fluff on your own data. I think we have a couple more minutes for questions. You can also me anytime if you have other questions and if you are interested in sending your thoughts on it more advanced plots you'd like to have information on, you can send suggestions. It's just a survey monkey, so just see what people are interested in in terms of advanced plots, but that is all I have. Do we have any questions? Justice Research and Statistics Association Webinar Page 13 of 15

14 Thank you Stephanie. Yeah, if anyone has any questions, please enter them into the chat box or you can raise your hand if you prefer to ask your question verbally. I did see that Jason went ahead and put a link in the chat box where we are going to be posting the webinar, so that you'll be able to access the data and start playing around in R. Great. Even if the questions are you couldn't get things into R, I'm happy to get that answered as well. While we still have people on the line here and while we're filling any questions, Jason go ahead and launch our poll. This is just our evaluation poll here, so go ahead and fill out the poll if you can or any chats for questions. Stephanie I had a question for you. Sure. I'm familiar with ArcGIS and I've done mapping using that software and similar to SPSS quite pricey. Seeing that R has the ability to create some GIS maps is great to see. I was curious to know you said that the dataset was using X and Y coordinates. Do I have the capability to use addresses or would you have to convert those prior to analyzing in R? There are files that you can download, so I'm going to combine this with maybe where I think you might be interested in going. You do not have to convert that to X, Y coordinates, that's just what this dataset actually had in it. They have plain locations to find, but if you have latitude, longitude, if you have GIS location information, there are resources where you can pull in map files to plot and then layer those on top of it. It actually does have quite a bit of capability with that as well. Great, thank you. I mean if you do a quick Google, you would be surprised that I say the sophistication of the plots that you can get to. It certainly is a utility for academics, but that also means all of their grandchildren that they like playing with and making super fancy, we love being able to share that and put that out there. There's a ton of resources for pretty sophisticated data visualization options. Excellent. Yeah, I've heard that that R has quite a community behind it and you really find a lot of help online which is great, especially for someone who might not be as familiar with the coding aspect. Mm-hmm (affirmative), absolutely. Okay. Well, I'm not seeing any chats here, but if you guys think of something afterward, feel free to Stephanie. Her is up there or you can Justice Research and Statistics Association Webinar Page 14 of 15

15 myself. It'd be of any questions that come up for you and like I said, the webinar is going to be posted on our website here in a couple days. Thank you everyone and thank you Stephanie so much for telling us about R. I've heard a lot of interest in this, especially the free aspect of it, so it's great. Absolutely. Please don't hesitate to reach out, even if it's a simple question, fine I'm happy to zoom with you and we can share screens. Really once you get past the first hurdle of using it, it allows you some very powerful options. Feel free to reach out or if you're shy, you don't want to do that and you just want to hit that survey monkey link at the bottom, I'm happy to take some suggestions down there as well so it's good to know what people are looking for. Great. Thanks again Stephanie and anyone else, just go ahead and finish the poll there if you can and everyone else have a great evening. Great, have a good one. Thanks. Justice Research and Statistics Association Webinar Page 15 of 15