Scatter Plot
Scatter PlotScatter plots allow you to visualize the variation in one variable relative to the variation in a second variable. In STIS, you can make two kinds of scatter plots: plots of any two datasets in the same geography (over multiple times), or plots comparing values in the same dataset from two different times. When you make a scatter plot, your output will consist of the set of locations from your focal geography plotted on an x and y axis based on their values for the two datasets or two times for the same dataset. A Local Moran analysis produces a related figure, a Moran scatter plot.
Like most views in STIS, scatter plots for two datasets can be animated; use the animation toolbar to scroll through the temporal range of your data. You can also synchronize the animation in two dataset scatter plots with animation in other views, such as maps. Both types of scatter plot views are by default linked to other visualization tools, so that areas selected in one plot, map, or table, are also selected in other visible views. For example, you may wish to select locations where values of both datasets are particularly high, and see where those values occur on the map.
|
Scatter plot for two datasets over a range of times, with graph statistics window shown (right side). |
Two times (or "timeless") scatter plot with graph statistic window hidden (this also hides the regression line). |
|
|
|
In addition to the plot, the scatter plot view includes a Graph Statistics
window that shows the mean and standard deviation of each data set at
the time shown in the animation
toolbar when you compare two datasets, or for the one focal dataset
at the two plotted times. This window also shows
two coefficients that measure the correlation between variables, the simple
correlation coefficient, and Spearman's rank correlation coefficient.
The correlation
coefficient page describes these measures in more detail. By
default, the statistics shown in this window are for all of the points
in the scatter plot, but you can select a subset of points with the cursor,
and then choose to show statistics calculated for this subset -- under
"Show statistics for:", click on the open circle next to "Selection."
You can hide 
or undock
the Graph Statistics window using the buttons in the upper right corner
of the window, and can bring back the window after hiding it by clicking
on "Graph Statistics" in the Graph pull-down menu.
When you activate the Graph statistics window, you will also activate the regression line within the plot. This simple linear regression line (also called the least squares regression line) represents the "best fit" line through your data, and is useful as a description of the relationship between two datasets, or for predicting values if you have a dataset that can be thought of as "dependent" on another dataset (an independent set, plotted on the x-axis). If you choose to show statistics for a selection, rather than the whole dataset, a second regression line will appear in orange. One way to use the selection option is to compare the original line to the one for the subset of data -- this will allow you to examine the influence of a few points on the relationship between to datasets. An example of this, as well as details on calculating the regression line, are presented on the regression line page.
The far left pull-down menu in the scatter plot window allows you to hide or show the Animation (shown by default) and Graph (hidden by default) toolbars. Note that if you wish to change the time step size for animations, this option is available from the Animation pull-down menu, but not from the toolbar.
From the Graph pull-down menu or toolbar, or from the right click menu, you have the following options:
You can alter the look of your scatter plot
by changing its properties.
Options here range from changes to the title and
way things are selected, to changes in the fill, size, and symbols used
for points. For two dataset plots, you can use
options within scatter plot properties to show variation in a covariate
data set. For example, the
plot image below illustrates the correlation between the log of the proportion
of hispanic females in western US counties and rates of cervical cancer,
with the log of proportion of population living below the poverty line
included as a covariate determining the size
of the plotted points.
You can print your scatter
plot.
You can export
an animated "two dataset" scatter plot as an .avi file.
You can copy
a scatterplot to the clipboard as an image file, and then paste it into
other software program files.
