Pertiwi, Violita (2018) Klasifikasi Microarray Kanker Prostat Menggunakan Metode Hybrid Support Vector Machine - Genetic Algorithm (Svm-Ga) Dan Naïve Bayes. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211645000020-Undergraduate_Theses.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Kanker merupakan masalah kesehatan yang cukup besar di seluruh dunia, salah satu jenis kanker yang banyak menyebabkan kematian adalah kanker prostat yang menduduki peringkat ke lima. Secara khusus, teknologi microarray telah diterapkan untuk prediksi dan diagnosis kanker, sehingga diharapkan dapat mendeteksi tumor atau kanker secara lebih dini dan tepat. Untuk mengklasifikasikan individu yang terjangkit tumor atau kanker secara tepat, pemilihan variabel yang berkaitan dengan kanker harus tepat karena ekspresi gen pada data microarray menghasilkan banyak gen. Pada data microarray jumlah variabel lebih besar dibandingkan jumlah observasi, sehingga perlu dilakukan klasifikasi dengan metode machine learning salah satunya metode Support Vector Machine (SVM) dan naïve bayes. Pada penelitian ini parameter SVM diatur menggunakan metode Grid Search, selain itu juga menggunakan Genetic Algorithm (GA) untuk mengoptimasi parameter SVM. Hasil analisis menunjukkan bahwa metode SVM-GA dengan seleksi variabel menggunakan FCBF menghasilkan performa klasifikasi yang lebih baik dibandingkan metode SVM dan naïve bayes untuk data microarray kanker prostat. Selain itu, optimasi parameter menggunakan metode GA dapat meningkatkan nilai akurasi pada klasifikasi data.
=============================================================================================Cancer is one of health problems in world. Prostate cancer is ranked fifth on causing death of human. In particular, microarray technology has been applied to prediction and diagnosis of cancer, so it is expected to detect tumors or cancers more early and precisely. In order to correctly classify individuals who are infected with a tumor or cancer, selection of variables related to cancer should be appropriate. Usually in microarray data, the number of variables is greater than the number of observations, so it is needed to be classified by machine learning method. One of suitable methods is Support Vector Machine (SVM) and naïve bayes. SVM is a machine learning that has been successfully used to solve classification problems in various fields. The problem in SVM is the difficulty of determining the optimal SVM parameter. In this research, SVM parameter is set using Grid Search method and Genetic Algorithm (GA) is used to optimize the parameter of SVM. GA is a population-based search that can seek a global optimum solution. The results of analysis show that the GA-SVM method gives better classification performance than SVM and naïve bayes for prostate data. In addition, parameter optimization using GA can improve the accuracy value in the data classification.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Genetic Algorithm, Microarray, Naïve Bayes, Optimasi Parameter, Support Vector Machine |
| Subjects: | H Social Sciences > HA Statistics |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Violita Pertiwi |
| Date Deposited: | 18 Jul 2021 23:03 |
| Last Modified: | 18 Jul 2021 23:03 |
| URI: | http://repository.its.ac.id/id/eprint/57536 |
Actions (login required)
 |
View Item |