Prediksi Financial Distress pada Perusahaan Asuransi Jiwa di Indonesia dengan Metode Klasifikasi dan Synthetic Features Generation

Purwanto, Dwi (2023) Prediksi Financial Distress pada Perusahaan Asuransi Jiwa di Indonesia dengan Metode Klasifikasi dan Synthetic Features Generation. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06211940000052-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (3MB) | Request a copy

Abstract

Permasalahan keuangan pada perusahaan asuransi jiwa menjadi masalah yang serius jika tidak segera ditangani. Laporan keuangan perusahaan harus selalu diawasi untuk mengantisipasi risiko terjadinya financial distress yang bisa menyebabkan kebangkrutan. Perusahaan dapat dikategorikan mengalami financial distress ketika memiliki rasio RBC kurang dari 120% atau ROA < 0, maka dari itu dilakukan prediksi financial distress untuk menilai kondisi keuangan terkini dari perusahaan sehingga dapat diambil langkah penyelesaian jika telah mengarah ke financial distress. Pada penelitian ini prediksi financial distress menggunakan metode klasifikasi Support Vector Machine (SVM), Logistic Regression, Generalized Extreme Value Regression (GEVR), dan Extreme Gradient Boosting (XGB) dengan melibatkan synthetic features generation secara serentak dan seleksi variabel. Hasil prediksi financial distress diperoleh model terbaik yang mampu memprediksi kondisi keuangan perusahaan asuransi jiwa di Indonesia pada setiap size dimana untuk size 0 adalah model XGB secara serentak dan melibatkan synthetic features generation dengan nilai akurasi 98,00%, AUC 100%, dan recall 100%. Kemudian, pada size 1 diperoleh model XGB secara seleksi variabel tanpa melibatkan synthetic features generation dengan nilai akurasi 94,10%, AUC 93,90%, da recall 100%. Lalu, untuk size 2 diperoleh model Stepwise GEVR dengan melibatkan synthetic features generation dengan nilai akurasi 90,20%, AUC 82,60% dan recall 50% sedangkan pada size 3 diperoleh model XGB secara seleksi variabel dengan melibatkan synthetic features generation dengan nilai akurasi 92,10%, AUC 100% dan recall 100%. Penambahan variabel synthetic features generation dan model Extreme Gradient Boosting cenderung mampu meningkatkan performansi model di beberapa model. Semakin tinggi size yang digunakan cenderung menurunkan performansi dari beberapa model sehingga tidak bisa dibandingkan antar size
====================================================================================================================================
Financial problems in life insurance companies are a serious problem if not addressed immediately. The company's financial statements must always be monitored to anticipate the risk of financial distress that can cause bankruptcy. The company can be categorized as experiencing financial distress when it has an RBC ratio of less than 120% or ROA < 0, therefore financial distress prediction is carried out to assess the current financial condition of the company so that settlement steps can be taken if it has led to financial distress. In this study, financial distress prediction uses the classification methods of Support Vector Machine (SVM), Logistic Regression, Generalized Extreme Value Regression (GEVR), and Extreme Gradient Boosting (XGB) by involving synthetic features generation simultaneously and variable selection. The results of financial distress prediction obtained the best model that can predict the financial condition of life insurance companies in Indonesia at each size where for size 0 is the XGB model simultaneously and involves synthetic features generation with an accuracy value of 98.00%, AUC 100%, and recall 100%. Then, at size 1, the XGB model is obtained by variable selection without involving synthetic features generation with an accuracy value of 94.10%, AUC 93.90%, and recall 100%. Then, for size 2, the Stepwise GEVR model is obtained by involving synthetic features generation with an accuracy value of 90.20%, AUC 82.60% and recall 50% while at size 3 the XGB model is obtained by variable selection by involving synthetic features generation with an accuracy value of 92.10%, AUC 100% and recall 100%. The addition of synthetic features generation variables and Extreme Gradient Boosting models tend to be able to improve model performance in some models. The higher the size used tends to reduce the performance of some models so that it cannot be compared between sizes

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Asuransi Jiwa, Extreme Gradient Boosting, Financial Distress, Generalized Extreme Value Regression, Synthetic Features Generation; Extreme Gradient Boosting, Financial Distress, Generalized Extreme Value Regression, Life Insurance, Synthetic Features Generation.
Subjects:	H Social Sciences > HG Finance > HG4028.V3 Valuation. Economic value H Social Sciences > HG Finance > HG8771 Life insurance Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Dwi Purwanto
Date Deposited:	23 Aug 2023 06:02
Last Modified:	23 Aug 2023 06:02
URI:	http://repository.its.ac.id/id/eprint/104351

Actions (login required)

View Item