This example shows how to choose your perfect car using LIONoso's data mining and visualization.
You have several car models with different characteristics.
The problem is multi-objective – no single solution satisfies everybody.
Individual preferences must be matched with existing possibilities and
the proper choice is the result of personal compromises based on knowledge.
The use case is a toy example, you can easily substitute cars with your
set of options for making a proper and informed decision.
LIONoso sample visualization(s): histogram
A histogram displays tabulated frequencies, shown as bars.
It shows the proportion of cases that fall in each category.
One immediately identifies how the different attributes (engine size, price, fuel consumption, ...)
are distributed and gets a first idea about the space of possibilities.
LIONoso sample visualization(s): bubble-chart and parallel coordinates
More details can be visualized with a colored bubble-chart (price, horse power, weight and engine size in this case).
The attributes to be shown as X and Y coordinates, color and size of the bubble can be immediately changed
with a couple of clicks. Below a second visualization (parallel coordinates) is
with the bubble-chart. When a bubble is clicked, the values of all
attributes are shown as a gray line in the parallel coordinates display.
If filters are active in the parallel coordinates display, only the filtered
cases appear in the bubble-chart, etc. Trying the effectiveness of seeing the same data
with synchronized views is much faster than explaining.
LIONoso sample visualization(s): navigation, clustering and parallel coordinates
Clustering is a way to group different cases together. In this case vehicles are grouped
according to the given characteristics. After grouping, one prototype case
for each group
can be visualized. This is a very effective way to compress the information and concentrate
on a relevant subset of possibilities, to then go back to the details once the favourite
cluster (group) has been identified. A parallel coordinate display can be present to filter
on each coordinate.
Download the LIONoso-ready data file: cars.lion