Garuda - Garba Rujukan Digital

Prediksi Arah Harga Cryptocurrency Menggunakan Hybrid Lstm Encoder dan Xgboost Head: Implementasi dan Evaluasi pada 10 Aset Digital Utama Ade Kurniawan; Muliyono; Hari Suriadi
Jurnal Ilmu Sosial, Ekonomi dan Pendidikan Vol. 1 No. 2 (2025): Oktober 2025
Publisher : Suria Academic Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67467/jisep.v1i2.43

The primary challenge in cryptocurrency price prediction lies in the market’s highly volatile, non-linear, and near–random walk behavior, which makes traditional predictive models unable to achieve consistent accuracy. This study aims to develop and evaluate a hybrid model combining Long Short-Term Memory (LSTM) and XGBoost to predict price direction and returns for ten major cryptocurrencies using daily data from 2023 to 2025. Historical data were processed through feature engineering, normalization, and sliding-window sequence construction, and the models were evaluated using TimeSeriesSplit to prevent data leakage. The results show that the hybrid model consistently outperformed both LSTM and XGBoost, achieving an average directional accuracy of 58.6%, significantly higher than the baselines (51.7% for LSTM and 53.6% for XGBoost). The average RMSE of 0.0289 indicates stable return predictions without systematic bias. Statistical validation through paired t-tests and McNemar tests confirmed the significance of the improvement at p < 0.001. A trading simulation using a 1-day holding period produced an annualized return of 41.5% with a Sharpe ratio of 1.12, outperforming the buy-and-hold strategy. These findings highlight that integrating LSTM’s temporal representation with XGBoost’s non-linear learning capabilities is an effective and computationally efficient approach for cryptocurrency price forecasting, offering practical value for the development of algorithmic trading systems.

Development of Contextual Teaching and Learning (CTL)–Based Digital Instructional Materials Integrated with Islamic Values for Pancasila Education at SDIT Raisya Andre Pandoe; Arwin; Hari Suriadi
Islamic Studies and Education Journal Vol. 1 No. 1 (2025): Edisi December 2025
Publisher : Suria Academic Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67467/isej.v1i1.56

Learning of Pancasila Education in elementary schools remains insufficiently contextual and lacks the utilization of digital media; therefore, it is necessary to develop CTL-based teaching materials integrated with Islamic values to foster students’ national and religious character. This study aims to develop and examine the feasibility of Digital Teaching Materials for Pancasila Education based on the Contextual Teaching and Learning (CTL) approach integrated with Islamic values for Grade IV students at Integrated Islamic Elementary Schools (SDIT). The research employed a Research and Development (R&D) method using the ADDIE model, which consists of five stages: Analysis, Design, Development, Implementation, and Evaluation. The research subjects included teachers and Grade IV students from four SDIT in Padang City: SDIT Arafah, SD Islam Nibros, SDIT Cendekia Andalas, and SDIT Permata. Research instruments consisted of validation sheets for material experts, media experts, and language experts, as well as practicality questionnaires for teachers and students, and learning outcome tests (pretest–posttest). The results showed that the developed product was categorized as highly valid with an average score of 93%, highly practical with a score of 93.5%, and effective in improving students’ learning outcomes with an N-Gain score of 0.75 (high category). Learning using CTL-based digital teaching materials was proven to enhance student engagement, understanding, and character formation, particularly in instilling the values of unity, ukhuwah (brotherhood), ta’awun (mutual help), and amanah (responsibility). Thus, the CTL-Based Digital Teaching Materials for Pancasila Education Integrated with Islamic Values are declared valid, practical, and effective as an innovative instructional resource for Pancasila Education in integrated Islamic elementary schools. This product is expected to serve as a reference for developing contextual and character-based digital learning media in the era of the Kurikulum Merdeka.

Predicting Cryptocurrency Price Direction Using a Hybrid LSTM Encoder and XGBoost Head: Implementation and Evaluation on Ten Major Digital Assets Ade Kurniawan; Muliyono; Hari Suriadi
Journal of Emerging Economics and Business Vol. 1 No. 1 (2025): December
Publisher : Journal of Emerging Economics and Business

Show Abstract | Download Original | Original Source | Check in Google Scholar

The primary challenge in cryptocurrency price prediction lies in the market’s highly volatile, non-linear, and near–random walk behavior, which makes traditional predictive models unable to achieve consistent accuracy. This study aims to develop and evaluate a hybrid model combining Long Short-Term Memory (LSTM) and XGBoost to predict price direction and returns for ten major cryptocurrencies using daily data from 2023 to 2025. Historical data were processed through feature engineering, normalization, and sliding-window sequence construction, and the models were evaluated using TimeSeriesSplit to prevent data leakage. The results show that the hybrid model consistently outperformed both LSTM and XGBoost, achieving an average directional accuracy of 58.6%, significantly higher than the baselines (51.7% for LSTM and 53.6% for XGBoost). The average RMSE of 0.0289 indicates stable return predictions without systematic bias. Statistical validation through paired t-tests and McNemar tests confirmed the significance of the improvement at p < 0.001. A trading simulation using a 1-day holding period produced an annualized return of 41.5% with a Sharpe ratio of 1.12, outperforming the buy-and-hold strategy. These findings highlight that integrating LSTM’s temporal representation with XGBoost’s non-linear learning capabilities is an effective and computationally efficient approach for cryptocurrency price forecasting, offering practical value for the development of algorithmic trading systems.

Intellectual Capital, Corporate Social Responsibility and Financial Performance: Evidence from Indonesian Banks Hari Suriadi
Journal of Emerging Economics and Business Vol. 1 No. 1 (2025): December
Publisher : Journal of Emerging Economics and Business

Show Abstract | Download Original | Original Source | Check in Google Scholar

This study aims to examine the effect of intellectual capital and corporate social responsibility on the financial performance of banking companies listed on the Indonesia Stock Exchange during the period 2021–2024. This study employs a quantitative approach using secondary data from annual reports and sustainability reports of 18 listed banks, resulting in 72 firm-year observations. Financial performance is measured by Return on Assets (ROA), intellectual capital is proxied by the Value Added Intellectual Coefficient (VAIC™), and corporate social responsibility is measured using the CSR Disclosure Index based on GRI Standards. Data are analyzed using multiple linear regression after classical assumption tests. The results indicate that intellectual capital has a statistically significant negative effect on financial performance, while corporate social responsibility does not have a significant effect on ROA. Simultaneously, intellectual capital and CSR do not significantly explain variations in bank profitability during the observation period. This study extends the Resource-Based View and Legitimacy Theory by showing that the financial impact of intellectual capital and CSR is context-dependent and not always reflected in short-term accounting performance. The study is limited to accounting-based performance measures and a short observation period. The findings suggest that banks should manage intellectual capital and CSR from a long-term strategic perspective.

Problematika Karakter Generasi Muda di Era Digital: Analisis Kritis Terhadap Tantangan Moral dan Sosial di Era Teknologi Informasi Zulkifli; Hari Suriadi; Neni Sriwahyuni
Journal of Social, Educational and Religious Studies Vol. 1 No. 2 (2025): September 2025
Publisher : Suria Academic Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67467/jsers.v1i2.25

Transformasi digital yang masif telah membawa dampak signifikan terhadap pembentukan karakter generasi muda. Di satu sisi, era digital menawarkan akses informasi dan peluang kreativitas yang luas, namun di sisi lain, ia juga menimbulkan tantangan serius dalam hal nilai-nilai moral, integritas, dan identitas diri. Tulisan ini bertujuan untuk menganalisis secara kritis berbagai problematika karakter generasi muda yang muncul akibat pengaruh media digital, serta menelaah faktor penyebab dan strategi penguatan karakter yang relevan dengan tantangan zaman. Kajian ini menggunakan pendekatan kualitatif-deskriptif dengan metode studi pustaka, menganalisis berbagai sumber ilmiah dan data empiris yang relevan mengenai disorientasi nilai moral, kecanduan digital, budaya instan, krisis identitas, dan paparan konten negatif. Hasil kajian menunjukkan bahwa krisis karakter generasi muda disebabkan oleh lemahnya pendidikan karakter yang kontekstual, degradasi peran keluarga, tekanan sosial-psikologis digital, dan kurangnya regulasi konten media. Tulisan ini merekomendasikan strategi penguatan karakter berbasis literasi digital kritis, integrasi etika digital dalam kurikulum, revitalisasi keluarga sebagai agen nilai, serta penguatan kolaborasi lintas sektor. Temuan ini diharapkan dapat memberikan kontribusi teoretis dan praktis bagi pengembangan pendidikan karakter yang adaptif terhadap era disrupsi teknologi informasi.

Prediksi Arah Harga Cryptocurrency Menggunakan Hybrid Lstm Encoder dan Xgboost Head: Implementasi dan Evaluasi pada 10 Aset Digital Utama Ade Kurniawan; Muliyono; Hari Suriadi
Jurnal Ilmu Sosial, Ekonomi dan Pendidikan Vol. 1 No. 2 (2025): Oktober 2025
Publisher : Suria Academic Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67467/jisep.v1i2.43

The primary challenge in cryptocurrency price prediction lies in the market’s highly volatile, non-linear, and near–random walk behavior, which makes traditional predictive models unable to achieve consistent accuracy. This study aims to develop and evaluate a hybrid model combining Long Short-Term Memory (LSTM) and XGBoost to predict price direction and returns for ten major cryptocurrencies using daily data from 2023 to 2025. Historical data were processed through feature engineering, normalization, and sliding-window sequence construction, and the models were evaluated using TimeSeriesSplit to prevent data leakage. The results show that the hybrid model consistently outperformed both LSTM and XGBoost, achieving an average directional accuracy of 58.6%, significantly higher than the baselines (51.7% for LSTM and 53.6% for XGBoost). The average RMSE of 0.0289 indicates stable return predictions without systematic bias. Statistical validation through paired t-tests and McNemar tests confirmed the significance of the improvement at p < 0.001. A trading simulation using a 1-day holding period produced an annualized return of 41.5% with a Sharpe ratio of 1.12, outperforming the buy-and-hold strategy. These findings highlight that integrating LSTM’s temporal representation with XGBoost’s non-linear learning capabilities is an effective and computationally efficient approach for cryptocurrency price forecasting, offering practical value for the development of algorithmic trading systems.

Title

Found 6 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 6 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 6 Documents
Search