Hi, if your first time user of JMP there's
a number of nice videos on YouTube I might suggest you
check out the SAS software channel to find JMP software basics for
professors and students it's a nice overview. Let me provide you with a basic JMP tour myself. I'm going
to go into JMP off my shortcut here on my desktop. What
pops up is the tip of the day you can always uncheck that off for
future entering of JMP. If you're a first
time user you might want to check out enter beginners tutorial. Now as you
suspect to open a file you can go to File then Open. JMP remembers the last folder that I
entered in this case it was a folder as we can see here of a bunch of JMP files. I
know these are JMP files because the little blue icon next to the file names. If you can see closely is a little
guy jumping that's indeed no pun intended the JMP symbol.
Let's check out another folder of mine. In this folder, I have not JMP files but you can see the Excel symbol. I have CSV files, comma separated value files, and Microsoft Excel worksheets. Here are
files that JMP can also recognize. Let me go ahead and open one of them. I opened house prices which are prices of homes relative to
the size square footage, number of bedrooms, number of bathrooms, offers and information on whether or not the
house is brick or not and what neighborhood it comes from. There are three
neighborhoods: east, north, and west. JMP recognizes data to be one of three
data types: "Continuous" those are data that you can number crunch, "Ordinal" that is data which might have numerics but there's really
no meaning behind the numeric other them for ranking purposes like "1" for
good health and "2" for poor health and then three "Nominal" or
sometimes called categorical which indicates what category does an observation fall in. In this case, brick or neighborhood. JMP tells you that it recognizes the
data type as one of those three by a symbol right next to the column names
here in the column panel. Here, we see a blue triangle if we click on that indeed JMP says you're dealing here
with continuous data. If you click on the brick symbol which is a little red bar chart, it shows you that brick is recognizing
it to be nominal or categorical you can't even choose continuous
because these aren't after all numeric data Now, a lot of our action will be under Analyze and Graph. Let me show you one option under Analyze, namely,
Distribution. I go ahead and click in a column. You can also drag it in. I get a histogram of the data along with some number summaries: quantiles and summary statistics. Now, you'll notice that the histogram is vertically laid out. This is a
little awkward to look at. In fact if you look on our lecture slides
you'll find the histograms are horizontally laid out. To look at it from the same perspective
you'd have to sort of turn your head to the left, look from the
backside. It's an awkward perspective in my personal opinion. We'll want to
change that. You'll see that we have next to the headers either a gray
triangle or a red triangle. The gray triangles have a couple purposes.
One of the most basic ones is just to close things out. If I don't want to look at that I can
close it out. I could always open it up. The red triangles
provides you with a number of options that you can do for things below. Here I click on the red triangle
and see that there's a number of options I could supplement the histogram with, for example, if I want
a normal quantile plot, it appends a normal quantile plot to the
histogram. Now, back to the issue of layout. If I want to lay this out
horizontally I would go to the upper red triangle and
choose the options from it and see that there's an option of
Stack. You could see here Stack puts it out horizontally. Now that's a bit of a pain to
have to go to Stack each and every time. We'll want to make this a default. I'll get to that in a moment. As I said
there's also a lot of stuff to be done under Graph.
Many graphing opportunities, let me go to one of the all-purpose ones of Overlay Plot. Overlay Plot allows us to
plot data as a time series or plot Y versus X plots. Let's plot, for example, price against square footage and you can see
here we have a scatterplot. The dots on the scatterplot are quite small.
We can change that. The dots are black. We can change that. Let me indeed show you how. We can right click on the region and you can see there's a
number of options here that we can pursue, for example, Marker Size. If I want to go with a larger marker size, we see now that all the dots have larger
marker size that might be easier to look at. We can also under Customize choose the Marker option. We can change
the marker. Let's make them solid squares. We can change the color. Let's make them red and there you see the changing of the graph. You can also
click on the axes to change the axis labels, mins and maxs. You can change the scaling from linear to log. Lots of things that you can do
with these graphs. Now, let me go back to the issue of default on vertical layout versus a
horizontal layout. You can go under File to Preferences. Here are all the what's called the manufacturing settings of JMP. You can change these defaults. I
would suggest you leave most of them alone. The ones that you might be most
interested in pursuing Graphs. Here under Graphs you see the
default is for a small graph marker which we had to change earlier. If you like it to be larger, let's make it Large each and every time you
can choose that as a default and I clicked Apply. If you go to Platforms, you can scroll down and find Distribution. That's where the histogram resides and
you'll see what shows is an Options box. You'll see
there's a number of options, some are checked off, some are not. The ones
that are checked off will automatically show up each time you go into the platform. The ones that are not
checked off you can add on later. They're the
ones that you can find under the red triangle. For example, we didn't get a
normal quantile plot when we got the histogram but we could
always add it later as we did earlier. You'll notice that there's the Stack
option which is unchecked. There's also a Horizontal
Layout. You can choose either one of them to give you a
horizontal layout interestingly enough. I'm just going to play it safe and
choose both and click OK. Now when I go to Analyze, Distribution, click in Price. I'm also going to click in another one to show
you a neat interaction that JMP can do. Click in neighborhood which is a
categorical or nominal variable but JMP will recognize how to deal with
it in terms of a distribution. We first see that now the histogram is laid horizontally
and that's great so we can leave that as a default for all future histograms. We see that the categorical variable
neighborhood is plotted and I think you could figure
out what it's doing here. There are 128 houses in this data set, 45 are from the East neighborhood, 44 are from the North, 39 are from the
West. So, it counts basically the number that fall in each one of
the categories and then it gives you the percentages or proportions out of the
128. Now, one neat thing you can do in JMP is that it's very interactive with its graphics. If you were to click on a point or a collection of points on one graph,
JMP will show you where they are on another graph. So, for example, suppose
I were to click North, you see a few things happen. First, on the price distribution you can
see where the houses that are North fall on that price distribution, looking here they tend to look and they're the
darker shaded values, they tend to be on the lower end to the
distribution. You can also notice that we have here in the data table those values highlighted, namely, the North values cause that's what I
clicked so if I wanted to change that, let's go
ahead and click for example West now the highlighted values in the data
table or all the houses that are from the West neighborhood. We can see from this that West
neighborhood homes are on the higher end to the distribution so this is really a nice way to do sort of informal exploratory data analysis visually to get a feel for
where data sort of in a correlative sense reside on one graph versus other graph. Now, if I click out you'll see that JMP
continues to have those rows selected, in particular, I left off on having West selected. If you look down here under the Rows panel, you'll see that there are 39 selected
out at 128. If you don't unselect
them they will continue to be selected in any subsequent analysis so if I were to
for example go back to graph and plot price per square footage again, you could see now the markets are larger
because we made that the default. You'll notice that there are a number points 39 in particular that are dark, those are the highlighted points
and the rest of the data are faded and sort of put into the background. Now, if we
don't want this to be the case anymore and we want all the points to appear we have to go to Selected and have them unselected. Now,
there's two ways actually to do it you can click on this (Selected) or you can come up here to Rows choose Rows and Clear Row States. Let me do it that way now you see that all those rows that
were highlighted are no longer highlighted. If I go back to the hist (excuse
me to the scatterplot), you see now that all the points are indeed dark and none of them are faded. I
mention this and I want to end on this quickly because sometimes when I've noticed for myself and for my students. You're navigating around you may
accidentally hit a point here ... let's say I'm wondering what this point is.
It's point 41 and you could see its been highlighted now as I said you may have accidentally hit
that point or you may be interested in investigating upon that point but
you'll see what happened: namely, that point is darkened and
then the rest that the scatter plot is faded. What I've had encountered is that
students would copy this graph let's say into Word and then print it out
and submit as part of a project I'd look at the
graph and say: "Boy, this graph is a little hard to look at" because indeed the points have been faded out.
And, this is probably an accident that was unknowingly done. So, be on
the alert if you have a graph where all the points are not dark and some of points or many of
points are faded out. You've probably have a row that been selected and you'll want to clear it out and so if
I go ahead and clear it out I'll hit on Selected (double click on it). Now, we get a scatterplot showing all the
points darkened out. Well, that's good for this video. We'll
will be revisiting you other issues in the future.
