After going through all the interaction methods and all the methods that are available to create multiple linked views, now I want to conclude with giving you a series of scenarios or situations that often arise in the design on of interactive visualization systems. This is by no means exhaustive, but it's a series of scenarios that when you encounter them, you can recognize them and you can also start thinking about possible solutions in terms of multiple views techniques that can be applied. So, let's start with the first scenario. The first one is a situation where you have a visualization that contains too much data. Because of that, it can't fit entirely in one single view. That's a very common problem. Every time you're trying to visualize somewhat large datasets, and the size of your visualization grows with the number of data points that you want to visualize. Then you have this problem. Let me see if I have an example. So, a good example is matrix visualization. The more rows or columns you have, the harder it is to feed the old matrix into one single view. Now, why is this a problem? Why is it a problem for a visualization too big? Well, in principle, what you can do is to just apply scrolling or panning. So, at any given time, you're only seeing a portion of your visualization. Then by scrolling or panning, you can see the areas that you're interested in. But the problem with this is two fault. The first one is that this way it's very hard if not impossible to gain an overview of your data, and getting an overview of your data may be really really useful, because you may spot interesting or important trends. Related to that, if your visualization doesn't fit and you have to scroll and pan, then making comparisons between areas of the visualization that can fit all at once in one single screen becomes really really hard. Why is it hard? Because you have to scroll, see something, scroll back to the original point, and compare it mainly using your memory. You can no longer use your vision. You have to use your memory, your visual memory in order to make comparison. This is much much less effective than just seeing it with your eyes. So, when the situation arises, a very common pattern that you can use is the overview plus detail pattern. What is the idea here? The idea here is that you can use two views. One view is an overview of the whole dataset, with a much much lower resolution. So, this way, you won't be able to have all the details of your visualization, but at least you will be able to see the major patterns. That's what you have in one view. Whereas in the other view, you can see the details of a region that is selected in the overview view. Typically, this is interactive. So, you can move a window in the overview part of your interface, and as you move this window, you see the details of this region in the detail view. This is an extremely powerful method, and it's used extensively in visual interactive interfaces and in user interfaces in general. So, there are some common configurations of this pattern. So, say for instance, you can arrange the view horizontally or vertically. So, this means that your window can only go up or down. In a way, a scroll bar is one example of an overview plus detail visualization. The scroll bar gives you information about in which position we are now, and the main view is the actually the detail of this position. The same thing can be done in 2D, and this is commonly used in maps. So, you have one view that shows the overview of the map, and then you have a window and you can move this window and you have the other view that shows all the details of the position that is currently selected in the overview. So, that's a very common pattern. Keep it in mind. Every time you have the problem that the old visualization with the all amount of details doesn't fit in one view, this solution is possible.
