Finally, graphs can display large data sets in a way that can be readily

perceived and understood.

A table with values becomes very difficult to display and understand.

The same is not true for graphs.

In fact, in the era of big data,

graphs sometimes are generated from millions or even billions of data points.

Many expert actually believes that

data visualization can be the last step of analytics project

if it is sufficiently displays the interesting aspect of the data.

To further contrast tables and graphs, it is helpful to

think about several scenarios for the sales data example we discussed before.

What if there are sales data for multiple products that we need to display?

This typical is not a problem for tables.

We can simply add a few columns to the table.

We may also do the same with graphs.

However, the task is trickier.

To start with, even one product will potentially need multiple graphs, which

means that information can be a little more scattered if we choose to use graphs.

Well, we can certainly display multiple products on each graph,

and it may not work well if the scale of the numbers is substantially different for

the different products.

What if sales for one product is in billions,

whereas the other one is only in millions?

That is not to say that tables are not always their own limitations.

If we have ten years of data, then we're going to have a very long table and

it is probably not feasible anymore.

Displaying the information graphically is only feasible method in that case.

A graph can also reveal patterns that are hard to discover in tables,

such as sales trend and seasonality.

Text, graphs, and tables are not mutually exclusive.

We can often combine them.

For example, for

really large datasets we can use graphs to show patterns in the data, and

use tables for summary information, possibly on multiple measurements.

We can then use text to emphasize individual values or

explain complex patterns and ideas.