Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal International Journal Software Engineering and Computer Science (IJSECS) SIENNA
Andreas Perdana
Unknown Affiliation
Computer Science & IT Engineering

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Optimasi Hyperparameter Bi-Directional Long Short Term Memory Menggunakan Particle Swarm Optimization Untuk Prediksi Saham BBRI Made Arya Adi Yoga; Andreas Perdana
Sienna Vol 7 No 1 (2026): Sienna Volume 7 Nomor 1 Juli 2026
Publisher : LPPM Universitas Muhammadiyah Kotabumi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47637/sienna.v7i1.2320

Abstract

Pasar modal, khususnya saham sektor perbankan seperti PT Bank Rakyat Indonesia (Persero) Tbk. (BBRI), sering kali menunjukkan volatilitas tinggi, sehingga membuat prediksi harga saham menggunakan metode konvensional menjadi sangat menantang. Tujuan penelitian ini adalah untuk meningkatkan keakuratan prediksi harga saham BBRI dengan mengintegrasikan arsitektur BiDirectional Long Short-Term Memory (Bi-LSTM) dan algoritma Particle Swarm Optimization (PSO). Salah satu tantangan utama dalam penerapan deep learning adalah penentuan kombinasi hyperparameter yang tepat. Oleh karena itu, PSO digunakan untuk mencari nilai optimal bagi parameter seperti ukuran tersembunyi, kecepatan pembelajaran, dan tingkat putus sekolah. Hasil penelitian menunjukkan bahwa PSO berhasil mengidentifikasi parameter optimal dengan learning rate sebesar 0.01000 danhidden size 96, yang memungkinkan model mencapai konvergensi pada epoch ke-29. Performa model yang dihasilkan menunjukkan akurasi yang baik dengan nilai Mean Absolute Percentage Error (MAPE) sebesar 1,32% dan Root Mean Squared Error (RMSE) sebesar 68,44. Kesimpulan dari penelitian ini menunjukkan bahwa penggunaan PSO dalam optimasi memberikan hasil yang cukup signifikan meningkatkan kemampuan Bi-LSTM dalam memodelkan Menggambarkan pola data waktu yang rumit, model ini bisa menjadi bantuan yang baik bagi investor dalam membuat keputusan investasi di pasar modal.
Classification of Anemia Severity Using Random Forest and SHAP Analysis on Complete Blood Count Data Tira Julia Indah Sari; Andreas Perdana
International Journal Software Engineering and Computer Science (IJSECS) Vol. 6 No. 2 (2026): AUGUST 2026
Publisher : Lembaga Komunitas Informasi Teknologi Aceh (KITA), Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35870/ijsecs.v6i1.7199

Abstract

Anemia remains a major global public health challenge, requiring accurate severity classification to support early diagnosis and clinical decision-making, particularly in resource-limited settings. This study aimed to develop an accurate and interpretable multiclass model for classifying anemia severity according to the 2011 WHO hemoglobin thresholds. A Random Forest model combined with SHAP (SHapley Additive exPlanations) analysis was developed and evaluated using a Complete Blood Count (CBC) dataset comprising 364 samples categorized into four classes: Normal, Mild Anemia, Moderate Anemia, and Severe Anemia. Data preprocessing included Interquartile Range (IQR) capping, label encoding, and StandardScaler standardization. The model was optimized through Grid Search Cross-Validation using 432 hyperparameter combinations and five-fold Stratified K-Fold validation. On the test set, the model achieved an accuracy of 95.89%, macro-precision of 96.61%, macro-recall of 97.14%, and macro-F1-score of 96.85%. The Moderate and Severe Anemia classes achieved F1-scores of 100%, although the result for Severe Anemia should be interpreted cautiously because of the limited number of test samples. Mild Anemia was the most challenging class, particularly for samples near the hemoglobin classification thresholds. SHAP analysis identified HGB, PCV, and RBC as the most influential features. However, the SHAP results should be interpreted in light of the correlations among these variables and the use of HGB as the basis for WHO severity labeling. The findings indicate that Random Forest combined with SHAP can accurately reproduce anemia severity classifications based on the applied labeling criteria while providing interpretable predictions. Further external and clinical validation is required before the model can be considered for use in a clinical decision support system.
Co-Authors Made Arya Adi Yoga Tira Julia Indah Sari