0:03

Now, after we've gone through several examples of data abstraction,

you may ask yourself,

why is it useful to identify attribute types?

Why, as a visualization designer,

does it make me better equipped to create effective visualizations?

Well, the idea here is that,

knowing what kind of attribute an attribute is,

is going to give you guidance in selecting appropriate graphical visual representations.

And this is something that we are going to discuss in much more details,

in the next lesson.

But here, I want to give you a few examples so

that you can grasp the concept a little better.

I don't want this concept to be too abstract,

even if we're talking about data abstraction.

So the first one, is an example showing you that in a line chart,

a chart like the one that you are seeing in front of you right now,

it's inappropriate to use unordered attributes.

So, data attributes that are not either ordinal or quantitative,

that don't have an order.

Okay. So, in the first plot,

I'm showing something changing over time.

So, on the X axis,

I have time and on the Y axis,

I have quantity that is changing over time.

So this is an appropriate use of a line chart.

But the line chart on the right hand side,

tries to use exactly the same design,

but on the X axis,

I have a number of categories.

In this case, I have a number of

different burrows in the vehicle collisions data-set that we use previously.

Now, when you look at this chart,

you may think that what this chart is showing,

is either something changing over time,

because you are used to see a line chart when on the X axis you have time.

Or you may think, even worse,

if you realize that on the X axis you have categories,

you may think that these chart is actually showing you some useful trend.

But actually, there is no particular trend here,

because one important characteristic

of categorical attributes when they are mapped into a chart like this one,

is that they can be reordered in any way because there's no intrinsic order.

But if you reordered this chart,

you can get completely different patterns.

So the patterns, are actually not meaningful.

That's an example, think about what I just said.

All these reasoning is based on

the idea of knowing what kind of attributes we are talking about.

Let me give you another example.

If we are using a bar chart,

we know that bar charts can accommodate

information about categories and frequencies,

or statistics associated to these categories.

Another useful thing, is what they just said previously,

is that we know that categorical attributes,

when they are mapped in a chart like this one,

can be reordered in any way we like.

And here, I reordered this bar chart according to the values,

the frequencies or the statistic that is mapped to the height of the bar.

And often being able to reorder the bars of a bar chart, is very useful.

Because it allows us to more easily read the progression of values.

Okay. One thing that is useful to know when we have categorical attributes,

is that categorical attributes can be reordered.

Note that, this is not true if you have a similar chart where on the X axis,

you actually have ordinal or quantitative data,

you can no longer reorder the bars in

a bar chart or other graphical elements if you are using a different chart.

Another example. If in the data-set we have spatial attributes,

and let's say that more specifically,

we have attributes that describe geographical information,

we know that one visual representation that is available to us, is a map.

Now, as we will see in the future,

using a map is not necessarily always

the best solution when we want to visualize spatial data.

But for sure what we know,

is that if we have spatial information, a spatial attribute,

we know that using a spatial visual representation like a map,

is one of the available options.

It's very useful. Last one.

Say that we want to visualize information that comes

from an attribute that we know to be quantitative and diverging.

Okay. In this specific visual representation,

we have a heat map,

also called a matrix,

where we have two categorical attributes,

and at the intersection of these two categorical attributes,

we have a quantity that is actually diverging,

it's a diverging quantitative attribute.

Now, notice the difference between the first one and the second one.

In the first one on top,

I am showing the value of

these quantitative attributes through

a color scale that uses different levels of intensity,

but exactly the same color hue.

Hue is basically the name of the color,

or the type of color right.

Whereas, in the one below,

I'm using a different color scale.

In this color scale,

we have three main color types.

We have red, white and blue.

And the zero value is mapped to white.

Positive values are mapped to different color intensities of blue,

and negative values are mapped to different color intensities of red.

Now, try to compare these two and figure out which cells have negative values,

and which cells have positive values.

The one on top is very hard to distinguish between positive and negative,

whereas, the second one is very easy to distinguish between positive and negative values.

Once again, knowing the characteristics of an attribute, in this case,

knowing that it's a quantitative diverging attribute,

helps us make specific decisions on how to visualize these attributes.