PERBANDINGAN PERFORMA REGRESI LOGISTIK MULTINOMIAL DAN JARINGAN SARAF TIRUAN UMPAN MAJU YANG MENGGUNAKAN ALGORITMA LEVENBERG-MARQUARDT
Performance measures of a model for classification problem in the case of the response variables more than two categories, can be seen from the sensitivity, specificity, and accuracy of classification. Which is considered the best method is a method that gives the smallest classification error. In o...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Theses and Dissertations NonPeerReviewed |
| Published: |
[Yogyakarta] : Universitas Gadjah Mada
2012
|
| Subjects: | |
| Online Access: | https://repository.ugm.ac.id/100983/ http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=57232 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Performance measures of a model for classification problem in the case of
the response variables more than two categories, can be seen from the sensitivity, specificity, and accuracy of classification. Which is considered the best method is a method that gives the smallest classification error. In other words, has the highest performance measures.
In this study, application of the comparative performance of both methods
performed on two sets of data, namely data on thyroid disease and simulated data. In the model of feed-forward neural network used backpropagation algorithm based on Levenberg-Marquardt (trainlm). Artificial neural network architecture is optimum in the thyroid data is a model (21-4-3), while the simulation data is a model (2-10-3). The results indicate that the sensitivity, specificity and accuracy of classification using feed-forward neural network has a higher value compared to using multinomial logistic regression. However, to ensure the results of the comparison is true, do resampling 10 times and recorded the results of each performance measure. After resampling, feed-forward neural network capable of delivering higher performance measures than the multinomial logistic regression model, both on the training and testing data. |
|---|