Information Technology Reference
In-Depth Information
And thus, the non-linear SVM is interpreted in the way of the following four
inductive rules (IF-THEN) that will be easy to understand:
IF (A19
>
0.617975) THEN class=7
IF (A19
≤
0.617975) AND (A2
≤
0.634350) THEN class
=7
≤
0.617975) AND (A2
>
0.634350) AND (A10
≤
IF (A19
0.163654) THEN
class=7
IF (A19
≤
0.617975) AND (A2
>
0.634350) AND (A10
>
0.163654) THEN
class
=7
Fig. 5.5.
Visualization of the decision tree explaining the SVM result with the Segment
dataset
5.5
Visualization Tools for Explaining SVM Results
We have studied some ways to try to explain SVM results by using graphical
representation of high dimensional data. The information visualization methods
guide the user towards the most appropriate visualizations for viewing mining
results (post-processing step). There are many possibilities to visualize data by
using different visualization methods, but all of them have strengths and weak-
nesses. We use the linking technique to combine different visualization methods
to overcome the single one. The same information is displayed in different views
with different visualization techniques providing useful information to the user.
