PENERAPAN MODEL MACHINE LEARNING DALAM PENILAIAN BISNIS PENDEKATAN PASAR: STUDI KOMPARATIF MULTIPLE YANG OPTIMAL DALAM MENGESTIMASI NILAI

SAPUTRO, DIKY (2025) PENERAPAN MODEL MACHINE LEARNING DALAM PENILAIAN BISNIS PENDEKATAN PASAR: STUDI KOMPARATIF MULTIPLE YANG OPTIMAL DALAM MENGESTIMASI NILAI. Skripsi thesis, Politeknik Keuangan Negara STAN.

	Text (Cover) 01. Cover_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (136kB)
	Text (Abstrak) 02. Abstrak_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (117kB)
	Text (Daftar Isi) 03. Daftar Isi_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (147kB)
	Text (Bab I) 05. Bab I_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (230kB)
	Text (Bab II) 06. Bab II_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (424kB)
	Text (Bab III) 07. Bab III_Diky Saputro_4112230008.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (209kB)
	Text (Bab IV) 08. Bab IV_Diky Saputro_4112230008.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (764kB)
	Text (Bab V) 09_Bab IV_Diky Saputro_4112230008.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (66kB)
	Text (Daftar Pustaka) 10. Daftar Pustaka_Diky Saputro_4112230008.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (160kB)

Abstract

Abstrak Penilaian bisnis memiliki peran penting dalam optimalisasi penerimaan negara melalui ekstensifikasi, pengawasan, dan pemeriksaan perpajakan. Namun, metode yang digunakan sering kali bersifat subjektif, terutama dalam pendekatan pasar, di mana pemilihan perusahaan pembanding dan multiple yang digunakan dapat menghasilkan variasi hasil penilaian yang berpotensi menimbulkan sengketa. Oleh karena itu, penelitian ini bertujuan untuk mengurangi subjektivitas dengan memanfaatkan model machine learning serta mengidentifikasi multiple yang paling akurat dalam pendekatan pasar. Studi ini menggunakan data perusahaan makanan dan minuman yang terdaftar di BEI selama 2014–2023. Data keuangan dianalisis menggunakan algoritma decision tree, gradient boosting, neural network, dan random forest. Hasil penelitian menunjukkan bahwa algoritma random forest memiliki performa terbaik dalam memprediksi valuasi dengan R² sebesar 82,8% dan RMSE 0,824. Selain itu, multiple Price to Sales (P/S) menunjukkan performa lebih baik dibandingkan Price to Book Value (P/B) dan Price to Earnings (P/E). Temuan ini memberikan rekomendasi berbasis data bagi penilai untuk meningkatkan objektivitas dan mengurangi subjektivitas dalam metode tradisional, sehingga menghasilkan penilaian yang lebih akurat dan konsisten. Abstract Business valuation plays a crucial role in optimizing state revenue through extensification, supervision, and tax audits. However, the methods used are often subjective, particularly in the market approach, where selecting comparable companies and valuation multiples can lead to varying results, potentially causing disputes. Therefore, this study aims to reduce subjectivity by utilizing machine learning models and identifying the most accurate multiple in the market approach. This study examines food and beverage companies listed on the Indonesia Stock Exchange (IDX) from 2014 to 2023. Financial data were analyzed using decision tree, gradient boosting, neural network, and random forest algorithms. The results indicate that the random forest algorithm performs best in predicting valuation, achieving an R² of 82.8% and an RMSE of 0.824. Additionally, the Price to Sales (P/S) multiple outperforms Price to Book Value (P/B) and Price to Earnings (P/E). These findings provide data-driven recommendations for appraisers to enhance objectivity and reduce subjectivity in traditional valuation methods, ensuring more accurate and consistent valuations. Keywords: Valuatin, Machine Learning, Market Aproach, Multiples

Actions (login required)

View Item