|
a |
|
b/DS_fusion.py |
|
|
1 |
""" |
|
|
2 |
Based on: |
|
|
3 |
"Data Classification Using the Dempster-Shafer Method. |
|
|
4 |
Qi Chen, Amanda Whitbrook, Uwe Aickelin and Chris Roadknight" |
|
|
5 |
|
|
|
6 |
Dempster’s rule of combination (DRC) |
|
|
7 |
""" |
|
|
8 |
|
|
|
9 |
# NOTE replace attributes for model of the ensemble |
|
|
10 |
|
|
|
11 |
# The training data is used to build the mass functions |
|
|
12 |
|
|
|
13 |
# 1 maximum and minimum values for each class based |
|
|
14 |
# on one attribute of the training data are found |
|
|
15 |
|
|
|
16 |
|
|
|
17 |
# The system combines the mass values from all four attributes using DRC, |
|
|
18 |
# thus producing overall mass values for all hypotheses, and the hypothesis |
|
|
19 |
# with the highest belief value is used to classify the data item. |
|
|
20 |
# If the hypothesis does not represent a single class, |
|
|
21 |
# then a second step is necessary. |