Rudiyanto, Charles (2019) Analisis Performansi SMOTE pada Klasifikasi Imbalance High Dimensional Data Berbasis Logistic Regression (Studi Kasus: Senyawa Obat Kanker). Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211540000030-Undergraduate_Theses.pdf Download (2MB) | Preview |
Abstract
Kanker merupakan salah satu penyakit penyebab kematian utama di seluruh dunia. Radioterapi adalah metode pengobatan menggunakan sinar pengion seperti sinar-X dan sinar gamma yang bertujuan untuk mematikan sel-sel kanker sebanyak mungkin dan memelihara jaringan sehat di sekitarnya. Radioterapi memiliki efek negatif, yakni dapat memperburuk kondisi pasien apabila jaringan normal disekitar sel kanker juga terkena paparan radiasi, termasuk p53 menginduksi apoptosis (kematian sel) jaringan dan sel normal. Radiasi membunuh sel-sel normal di sekitar sel-sel kanker. Dalam rangka menanggulangi efek radioterapi maka pada penelitian ini dilakukan analisis mengenai 84 komponen senyawa dengan 217 prediktor yang dapat menjadi proteksi radiasi atau radioprotector dengan melakukan dua percobaan yang dicobakan kepada sel normal serta sel yang terkena radiasi sinar gamma. Adapun metode yang akan digunakan pada penelitian ini yaitu Logistic Regression Ensemble (LORENS) dan Ensemble Logistic Regression (ELR) dengan Synthetic Minority Oversampling Techniuqe(SMOTE) dan tanpa SMOTE untuk mengklasifikasikan senyawa obat kanker untuk optimasi proteksi radiasi dan toksisitas yang dianggap baik sebagai radioprotector. Hasil analisis menunjukkan bahwa metode LORENS dengan 5 subruang threshold 0,5 menghasilkan nilai AUC terbaik sebesar 0,7.
==================================================================================================================================
Cancer is one of the leading causes of death throughout the world. Radiotherapy is a method of treatment using ionizing rays such as X-rays and gamma rays that aim to kill as many cancer cells as possible and maintain healthy tissue around them. Radiotherapy has a negative effect, which can aggravate the patient's condition if the normal tissue around the cancer cell is also exposed to radiation exposure, including p53 inducing normal tissue and cell apoptosis (cell death). Radiation kills normal cells around cancer cells. In order to overcome the effects of radiotherapy, this study analyzed 84 components of compounds with 217 predictors that could be radiation protection or radioprotector by conducting two experiments that were tested on normal cells and cells affected by gamma radiation. The methods to be used in the study this is the Logistic Regression Ensemble (LORENS) and Ensemble Logistic Regression (ELR) with Synthetic Minority Oversampling Technique (SMOTE) and without SMOTE to classify cancer drug compounds to optimize radiation protection and toxicity that is considered good as a radioprotector. The results of the analysis show that the LORENS method with 5 subspace 0.5 thresholds produces the best AUC value of 0.7.
| Item Type: | Thesis (Other) |
|---|---|
| Additional Information: | RSSt 519.53 Rud a-1 2019 |
| Uncontrolled Keywords: | ELR, LORENS, Proteksi Radiasi, SMOTE, Toksisitas |
| Subjects: | H Social Sciences > HA Statistics > HA29 Theory and method of social science statistics |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Charles Rudiyanto |
| Date Deposited: | 28 Aug 2023 05:16 |
| Last Modified: | 28 Aug 2023 05:16 |
| URI: | http://repository.its.ac.id/id/eprint/64068 |
Actions (login required)
 |
View Item |