HYBRID OF ADABOOST ALGORITHM AND NAÏVE BAYES CLASSIFIER ON SELECTION OF CONTRACEPTION METHODS
| Main Authors: | Faridah, Nurul, Dewi, Candra, Soebroto, Arief Andy |
|---|---|
| Format: | Article info application/pdf eJournal |
| Bahasa: | eng |
| Terbitan: |
Lembaga Penelitian dan Pengabdian kepada Masyarakat Universitas Brawijaya
, 2021
|
| Subjects: | |
| Online Access: |
https://jeest.ub.ac.id/index.php/jeest/article/view/231 https://jeest.ub.ac.id/index.php/jeest/article/view/231/212 https://jeest.ub.ac.id/index.php/jeest/article/downloadSuppFile/231/98 |
Daftar Isi:
- Stunting is a growth failure in children. Stunting can be avoided by adjusting birth spacing or implementing a Family Planning program by using appropriate contraception. Therefore, it is necessary to develop appropriate and rapid contraceptive selection techniques to assist family planning programs. This study develops a model for determining contraceptive methods using a Naïve Bayes Classifier. In addition, an Adaboost algorithm was used to handle the independent between attributes on Naïve Bayes. The performance evaluation of model was measured by combining k-fold cross validation and confusion matrix. Based on the results testing was obtained an optimal parameter of learning rate was 0.1 and the number of iterations was 50. The evaluation using optimal parameters produce the best accuracy of 87.5%, precision of 87.6%, recall of 87.5%, and f1-measure of 87.5%. This result was better than applying the Naïve Bayes without implementing Adaboost, which had 70% accuracy. The used of Adaboost was proven to increase the accuracy of Naive Bayes by 17.5%.