PENERAPAN QUICKPROPAGATION DAN ATTRIBUTE REDUCTION BERDASARKAN INFORMATION GAIN DAN DISCERNIBILITY MATRIX UNTUK PREDIKSI STROKE ULANGAN
The death can be caused by illness, one of them are because of reccurent stroke. Recurrent stroke can be predicted because there are some factors that raising the risk of occuring after having the first stroke. This reccurent stroke problem can be predicted thus to solving this problem a classifier...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Theses and Dissertations NonPeerReviewed |
| Published: |
[Yogyakarta] : Universitas Gadjah Mada
2014
|
| Subjects: | |
| Online Access: | https://repository.ugm.ac.id/131902/ http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=72413 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universitas Gadjah Mada |
| Summary: | The death can be caused by illness, one of them are because of reccurent
stroke. Recurrent stroke can be predicted because there are some factors that
raising the risk of occuring after having the first stroke. This reccurent stroke
problem can be predicted thus to solving this problem a classifier that able to
relate factors into right prediction is needed.
Artificial neural network with quickpropagation is an improved and faster
algorithm based on backpropagtaion. But there is one problem, function of an
artificial neural netwrok will grows complex if it's component is not on suitable
parameters including the dimension of input variables. There is a chance that a
variables from acquired stroke data is not relevan nor on good distribution. Neural
networks can be resulting with low accracy. To avoid this problem, taking relevant
subset from all variables is used which this method is called attribute reduction.
Using this method neural network can run better and get high accuracy on
predictioning reccurent stroke problem.
In order to predict reccurent stroke problem, sample data from this
reccurent stroke is needed. Sample data is obtained from RSUP Dr. Sardjito.
Sample data is processed beforehand so quickpropagation can do prediction. This
research run some test from sample data to get accuracy.
Result of this research shows that quickpropagation can predict reccurent
stroke problem with 74% accuracy without attribute reduction. Using attribute
reduction with high-entropy subset is resulting 74% accuracy and 76% accuracy
with discernibilty matrix.
Keyword : neural network, quickpropagation, attribute reduction, recurrent stroke |
|---|