IMPLEMENTASI ALGORITMA CATBOOST REGRESSOR DENGAN PENDEKATAN TARGET DIFFERENCING DAN INDIKATOR MAKROEKONOMI UNTUK PREDIKSI ARAH PERGERAKAN SAHAM PT BANK CENTRAL ASIA TBK

Arminto, Dito Imanuel and Luqman, Luqman (2026) IMPLEMENTASI ALGORITMA CATBOOST REGRESSOR DENGAN PENDEKATAN TARGET DIFFERENCING DAN INDIKATOR MAKROEKONOMI UNTUK PREDIKSI ARAH PERGERAKAN SAHAM PT BANK CENTRAL ASIA TBK. Diploma thesis, Institut Teknologi PLN.

[thumbnail of Cover_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
Cover_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (206kB)
[thumbnail of Lembar Pengesahan_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
Lembar Pengesahan_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (557kB)
[thumbnail of Abstrak_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
Abstrak_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (196kB)
[thumbnail of BAB I_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
BAB I_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (396kB)
[thumbnail of BAB II_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
BAB II_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (887kB)
[thumbnail of BAB III_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
BAB III_Skripsi_202231074_Dito Imanuel Arminto.pdf
Restricted to Registered users only

Download (1MB) | Request a copy
[thumbnail of BAB IV_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
BAB IV_Skripsi_202231074_Dito Imanuel Arminto.pdf
Restricted to Registered users only

Download (970kB) | Request a copy
[thumbnail of BAB V_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
BAB V_Skripsi_202231074_Dito Imanuel Arminto.pdf
Restricted to Registered users only

Download (216kB) | Request a copy
[thumbnail of Daftar Pustaka_Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
Daftar Pustaka_Skripsi_202231074_Dito Imanuel Arminto.pdf

Download (263kB)
[thumbnail of Skripsi_202231074_Dito Imanuel Arminto.pdf] Text
Skripsi_202231074_Dito Imanuel Arminto.pdf
Restricted to Registered users only

Download (4MB) | Request a copy

Abstract

Peramalan arah pergerakan harga saham merupakan tantangan fundamental dalam disiplin ilmu Data Science akibat karakteristik volatilitas pasar yang tinggi dan sifat data yang heavy-tailed. Penelitian ini bertujuan untuk membangun model prediksi arah tren saham PT Bank Central Asia Tbk (BBCA) dengan mengatasi keterbatasan metrik nominal dan risiko data leakage yang sering dijumpai pada riset terdahulu. Metodologi yang diusulkan mengadopsi kerangka kerja Knowledge Discovery in Databases (KDD) dengan mengimplementasikan algoritma CatBoost Regressor yang secara empiris terbukti tangguh menangani data tabular finansial.
Inovasi utama dalam penelitian ini mencakup penggunaan pendekatan Target Differencing untuk mengatasi limitasi ekstrapolasi model berbasis pohon, serta penerapan teknik Strict Lagging selama 20 hari bursa pada indikator makroekonomi (Inflasi, BI Rate, dan Kurs USD/IDR) guna menjaga integritas informasi dan meminimalisasi potensi bias masa depan (look-ahead bias). Dataset penelitian mencakup periode pengamatan dari Januari 2020 hingga Maret 2026 yang diekstraksi melalui API Yahoo Finance, Federal Reserve Economic Data (FRED), serta rekam historis kebijakan resmi Bank Indonesia.
Kinerja model dievaluasi melalui dua skenario komparatif (Ablation Study) dengan fokus utama pada metrik Directional Accuracy (DA) dan Confusion Matrix. Penelitian ini diharapkan memberikan validasi empiris mengenai efektivitas integrasi variabel makroekonomi ber-lag terhadap peningkatan akurasi arah prediksi. Hasil akhir sistem diproyeksikan sebagai basis Sistem Pendukung Keputusan (Decision Support System) yang dapat memberikan pertimbangan tambahan bagi pelaku pasar modal dalam mengantisipasi kecenderungan arah pergerakan aset.

Forecasting stock price direction remains a fundamental challenge in the field of Data Science due to the high volatility and heavy-tailed nature of market data. This research aims to build a predictive model for the price trend direction of PT Bank Central Asia Tbk (BBCA) stock by addressing the limitations of nominal metrics and the risk of data leakage commonly found in prior studies. The proposed methodology adopts the Knowledge Discovery in Databases (KDD) framework by implementing the CatBoost Regressor algorithm, which has been empirically proven robust in handling tabular financial data.
The main innovations in this research include the use of a Target Differencing approach to overcome the extrapolation limitations of tree-based models, as well as the application of a Strict Lagging technique of 20 trading days on macroeconomic indicators (Inflation, BI Rate, and USD/IDR Exchange Rate) to preserve information integrity and minimize potential look-ahead bias. The research dataset covers the observation period from January 2020 to March 2026, extracted through the Yahoo Finance API, Federal Reserve Economic Data (FRED), and official Bank Indonesia policy records.

Item Type: Thesis (Diploma)
Uncontrolled Keywords: CatBoost, Target Differencing, Strict Lagging, Directional Accuracy, Makroekonomi, Saham BBCA CatBoost, Target Differencing, Strict Lagging, Directional Accuracy, Macroeconomics, BBCA Stock
Subjects: Bidang Keilmuan > Algoritma
Bidang Keilmuan > Data Science
Bidang Keilmuan > Economics
Bidang Keilmuan > Finance
Bidang Keilmuan > Machine Learning
Bidang Keilmuan > Quantitative and Qualitative Methods
Skripsi
Bidang Keilmuan > Teknik Informatika
Divisions: Fakultas Telematika Energi > S1 Teknik Informatika
Depositing User: FAKULTAS TELEMATIKA ENERGI
Date Deposited: 24 Aug 2026 06:36
Last Modified: 21 Sep 2026 06:47
URI: https://repository.itpln.ac.id/id/eprint/7174

Actions (login required)

View Item
View Item