0:01
Okay, so let's cover some options for scatter plotting in plot_ly.
And it will do this just because scatter plots are such an obviously common thing
you might want to do.
So here is one thing you can change, the color of the points.
Color is as.factor(cyl).
So on the mtcars dataset.
So we come down here head(mtcars) and show you.
So cylinder is, where is it?
Cylinder is right there,
it's I think > table(mtcars$cyl).
Let's see yeah, so 4 6 and
8 so is displaying the number of cylinders of the car,
obviously it should have an impact on miles per gallon, which is our y variable.
So we might want to display that as another dimension
because it's effectively categorical, right?
It's just the three values.
We could do it as color, okay?
Now it's not by default, a factor variable, it's numeric.
So we are going to convert it to a factor variable by as.factor cylinder, and
now let's run it, and then you can just see now the colors represent cylinder and
up here is legend, by default it says 8, 6, and 4.
OK.
Now let's try something that's continuous to display the color.
So in this case if you get displacement, so heading back down here.
If you look at the mtcars dataset the displacement
is kind of a continuous variable, right?
So if we run that, plot_ly is smart enough to know
that the color is a continuous variable so it treats it as another dimension,
continuous dimension using a continuous color gradient.
And then it displays,
very nicely displays the key to the gradient here on the side, okay?
So that's another way you can use color.
1:59
Let's change the size of the points.
I think when we look at this plot, the points to me are very small.
And then as you saw maybe from the last lecture when I outputted it to an HTML
file, the points were almost invisible.
So let's change the size.
Now, in this case the size we're going to do is horsepower.
Okay? So, horsepower is a continuous variable so
the size will be every point is just going to have a different horsepower.
Okay, I'm sorry, every point is going to have a different size.
Okay?
So, here the cylinder is the color, right?
And horsepower is the size.
2:42
So another interesting way.
You run out of dimensions of course in a 2D scatter plot so
the only real dimensions you have are the two
displayed dimensions color, size of the points and then the plotting point it self
are the different kinds of dimensions you can display.
So here it’s nice that we're showing four dimensions miles per gallon,
weight, cylinders and horsepower all in the same plot.
3:49
a web based translation of the open GL graphics library.
Okay. And it's a way that
you can have interactive 3D graphics embedded in webpages.
So again this'll all just display and
be interactive like this when you output it as a webpage as well.
4:19
in this case we create a very fictitious data set.
So temperature was just 100 random normals,
pressure was just 100 random normals and dtime was just the numbers from 1 to 100.
So, there's a hundred data points, they have three different values, okay and
this just shows you plot_ly, same command, okay.
Now we just specify three variables, x, y, and z, so super easy.
type = "scatter3d" okay.
mode = "markers".
And this, the color = temp, just like before puts a color gradient.
On the points and then,
puts the actual key to the gradient on the side of the plot.
Okay. So,
that's all that we had to do to get this nice interactive 3D scatter plot.
5:14
Okay.
So, your Homework now.
So I feel like we've covered enough of scatter plots to get you started
on scatter plotting in plot_ly.
You could of course, look at the further options that you can try and play around
with, but so for homework what I'd like you to do is create a variety of different
2D scatter plots changing the plotting attributes, plotting other dimensions
using color, plotting other dimensions using different plotting points and so on.
And then try to do the same thing for a 3D scatter plot and
just again, output it a a web page so
you know how to look at it, you know how to work with these things as web pages.
5:56
And if you want, publish them to rpubs or something like that so
you have a hosting, a way to host it publicly.
Okay, so try that out and then next we'll just go over some different,
some even I think cooler applications of plot_ly