Analisis Ensemble Support Vector Machine dan Survival Support Vector Machine pada Data Nasabah Gadai di Perusahaan Financial Technology-X

Riyadi, Mohammad Alfan Alfian (2018) Analisis Ensemble Support Vector Machine dan Survival Support Vector Machine pada Data Nasabah Gadai di Perusahaan Financial Technology-X. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 06211650010027-Master_Thesis.pdf]
Preview
Text
06211650010027-Master_Thesis.pdf - Accepted Version

Download (2MB) | Preview

Abstract

Terdapat dua kategori nasabah gadai pada perusahaan Fintech X yakni nasabah early payment dan late payment. Setiap kategori nasabah terdapat durasi pelunasan barang tanggungan. Oleh sebab itu penting bagi perusahaan untuk mendapat informasi awal terkait kondisi nasabah apakah baik atau buruk. Nasabah yang baik adalah nasabah yang semakin cepat dalam melunasi tanggungan sedangkan nasabah yang buruk merupakan nasabah yang semakin lama melunasi tanggungan. Untuk mengatasi problem tersebut terdapat dua tahap permodelan yang dilakukan. Tahap pertama adalah klasifikasi nasabah yang early payment atau late payment. Tahap kedua menganalisis survival untuk masing-masing kategori nasabah. Adapun metode yang digunakan pada tahap pertama yakni Regresi Logistik Biner, SVM dan Ensemble SVM. Sedangkan pada tahap kedua adalah Cox Proportional Hazard dan survival SVM. Untuk mendukung kesimpulan pada tahap klasifikasi, dilakukan studi simulasi dengan membangkitan beberapa skenario variabel prediktor. Hasil studi simulasi diperoleh bahwa Ensemble SVM mampu mengimbangi kinerja SVM dan regresi logistik. Akan tetapi ketika diaplikasikan pada data nasabah Fintech X, peforma metode klasifikasi yang diajukan tidak memberikan hasil yang baik. Hal tersebut disebabkan tidak adanya variabel yang benar-benar dapat mendiskriminasi kategori nasabah early payment maupun late payment. Pada tahap berikutnya, survival SVM memiliki peforma yang baik dibandingkan Cox Proportional Hazard. Survival SVM unggul pada setiap kategori nasabah. Salah satu kemungkinan survival SVM unggul karena asumsi dari Cox Proportional Hazard tidak terpenuhi.
======================================================================================================
There are two categories of pawning customers in Fintech X companies, namely early payment and late payment customers. Each category of customer there is the duration of repayment of dependent goods. Therefore it is important for the company to get initial information related to the condition of the customer whether good or bad. A good customer is a customer who is getting faster in paying off the dependents while a bad customer is a customer who is paying off the dependent longer. To overcome the problem there are two stages of modeling. The first stage is the classification of customers who are early payment or late payment. The second phase analyzes survival for each customer category. The method used in the first stage of Binary Logistic Regression, SVM and Ensemble SVM. While in the second stage is Cox Proportional Hazard and SVM survival. To support the conclusions at the classification stage, a simulation study was conducted by generating some predictor variable scenarios. The results of the simulation study found that Ensemble SVM is able to compensate for SVM performance and logistic regression. However, when applied to customer data Fintech X, the performance of the proposed classification method does not give good results. This is due to the absence of variables that can really discriminate the category of early payment customers and late payment. In the next stage, SVM survival has a better performance than Cox Proportional Hazard. SVM Survival excels in every customer category. One possible survival of SVM better because the assumption of Cox Proportional Hazard is not met.

Item Type: Thesis (Masters)
Additional Information: RTSt 519.536 Riy a-1 3100018078368
Uncontrolled Keywords: Analisis Survival, Gadai, Cox Proportional Hazard, Ensemble Support Vector Machine, Survival Support Vector Machine
Subjects: Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
Divisions: Faculty of Mathematics, Computation, and Data Science > Statistics > 49101-(S2) Master Thesis
Depositing User: Mohammad Alfan Alfian Riyadi
Date Deposited: 15 Oct 2020 04:33
Last Modified: 15 Oct 2020 04:33
URI: http://repository.its.ac.id/id/eprint/55705

Actions (login required)

View Item View Item