0:12
Zooming is a kind of filter where remove extraneous data based on
the plotted coordinates of the data.
When we select a zoom window,
we chose to focus on some of the data based on its displayed coordinates and
to eliminate data outside of a window based on its displayed coordinates.
That's a kind of filter based on displayed coordinates, but a filter can be applied
to other attributes of the data besides the displayed coordinates.
And so in this case, we've got our plot of all the countries
of based on population and life expectancy.
We can filter out all the countries that are in other continents and
focus just on the countries that are in say, Africa.
And so by filtering on Africa, we only plot the countries in Africa and
the data is spread out nicely and there's much less data here.
So we've got room to be able to plot the names of the individual countries and
you can start to see the subset of the larger dataset that corresponds to Africa.
And you realize that many of the countries that have low life
expectancy are in Africa, but also there's a separate
band of countries in Africa that have higher life expectancies.
And by plotting the names, you've now got room to see more of the details,
because of this filter.
You can see that those countries are the ones located in Northern Africa and
there's different ways of selecting data by applying filters.
If you have ordinal or nominal data, for example, continent.
Then you can click a check box, you can show all of the different categories and
select the ones that you want to look at and the ones that you don't want to
be plotted or if you want to look at quantitative data.
For example, if we want to look at countries based on gross domestic product
or other quantitative fields, then we can set a range with a minimum and
a maximum value and filter out data above or below these extremal
values in order to focus on certain subsets of a quantitative range.
So in tableau, it's quite easy to filter data based on attributes,
you would just drag the attribute into this filter box.
And so in this case, if we have our countries displayed here
differentiated by putting the country in the marks box,
we may want to filter the countries based on continent.
And so, I dragged the Region dimension into the filter box and
that will filter based on regions.
And so now since region, these continents are basically a category,
they're presented to me in check boxes.
And so if I want to only look at a subset of the country say,
Africa, I can just click Africa and then click OK.
3:36
And then if we want to know what those countries are, then I can drag country
into the label box here and we can get labels for each one of these countries.
And you can see that the countries in Northern Africa tend to have
a higher life expectancy than the countries in the rest of Africa.
We don't want to overwhelm a display of a lot of information with further details,
especially when those details are extraneous and
distract us from general trends that we should be seeing in the data and
a very common way of doing this is with a tool tip.
If your mouse is sitting over a data item, then a little pop-up
window will come up and give you the detail based on that data.
And maybe some items that, like you can select only that item or
you can exclude that item and look at other operations based on that item.
A second way of finding details on demand is a field selection.
You can select a field of your dataset and
you can find the data points that field corresponds to or
a certain setting of that field corresponds to that get highlighted.
And so there's all sorts of interactive details that you can make available
to a user.
You just want to make them available only when they're requested and
not all at once, so as to not overwhelm the viewer.
5:07
So those are the basic steps for setting up a good
information visualization system to understand abstract information.
That first step, you want to set up an overview to see all the data.
The second step is you want to be able to drill down to the details of
that data by focusing, zooming, filtering.
And then that third step, you want to be able to provide details on your data, but
only when they're requested.
Only details on demand.
[MUSIC]
John C. HartProfessor of Computer Science
