Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Teuku Rizky Noviandy has a total of 157 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Investigating long-term impact of agricultural factors and renewable energy on environmental quality: Evidence from Indonesia using ARDL and DOLS

    by Ghalieb Mutig Idroes, Qalbin Salim Fazli, Iin Shabrina Hilal, Irsan Hardi, Teuku Rizky Noviandy, Dimas Bagus Wiranatakusuma


  • Open Access

    Development of an Interactive Web-Based Hybrid E-Commerce Recommendation System Using Neural Collaborative Filtering and Content-Based Filtering

    by Muhammad Farid, Tasyaul Husna, Muhammad Zia Ulhaq, Teuku Muhammad Faiz Nuzullah, Teuku Rizky Noviandy, Novi Reandy Sasmita

    In an increasingly digital era, technological advancements have transformed the way people shop and interact with e-commerce platforms. E-commerce offers a wide range of products and services online, but consumers often feel overwhelmed by the sheer number of choices. The importance of efficient and relevant services for consumers is a significant consideration in developing e-commerce systems. One effective solution is implementing a recommendation system that helps consumers find products that match their needs and preferences. This study develops an e-commerce recommendation system using neural collaborative filtering and content-based filtering methods, integrated with an interactive dashboard to improve user experience. With a learning rate of 0.01 and a batch size of 32, the developed neural collaborative filtering model shows a low testing error of 0.0705 and precision, recall, and F-measure of 0.9583, 0.8251, and 0.8867, respectively. The content-based filtering method uses the Cosine Similarity technique, achieving an Average Precision of 0.804 and an accuracy of 80%. The interactive dashboard was successfully developed to support system evaluation and user engagement. This study provides benefits for consumers who can find products according to their needs more quickly and for e-commerce business owners who can reach consumers more efficiently.

  • Open Access

    Optimizing Machine Learning Models for Cardiovascular Drug Candidate Screening with XGBoost and Successive Halving

    by Teuku Rizky Noviandy, Ryan Setiawan, Rahmat Sufri, Anisah Anisah

    Thrombin, a key enzyme in the blood coagulation process, is an important target in the development of anticoagulant therapies. This study proposes an efficient and accurate quantitative structure–activity relationship (QSAR) modeling framework that combines the XGBoost algorithm with the Successive Halving (SH) method for hyperparameter optimization. A dataset of 3,145 compounds was collected from the ChEMBL database, and molecular descriptors were generated using the Mordred calculator. After preprocessing and feature selection, the SH-tuned XGBoost model achieved the highest performance, with an accuracy of 85.69%, F1-score of 80.93%, and ROC-AUC of 0.917. Compared to baseline and Random Search-tuned models, the SH-tuned model demonstrated superior predictive accuracy and significantly reduced training time. Confusion matrix analysis confirmed the model’s strong sensitivity and balanced classification of active and inactive compounds. These results illustrate the effectiveness of combining gradient boosting with efficient hyperparameter optimization for virtual screening. The proposed framework reduces computational cost without compromising model quality and can support early-stage drug discovery. Future extensions may include regression modeling, expanded datasets, and integration of biological or pharmacokinetic data to further improve performance and applicability.


  • Top Cited Papers
    Open Access

    Investigating hydropower energy consumption's effect on Southeast Asia's path to achieving environmental sustainability and carbon neutrality

    by Ghalieb Mutig Idroes, Iffah Hafizah, Djoni Hartono, Dian Budi Dharma, Irsan Hardi, Teuku Rizky Noviandy, Rinaldi Idroes

    Hydropower presents significant environmental advantages over fossil fuels, particularly due to its minimal emissions during operation. This research examines the dynamic effects of hydropower energy consumption, fossil fuels, economic growth, manufacturing, agriculture, capital formation, and labor on CO2 emissions in Southeast Asian (SEA) countries from 2000 to 2022. Utilizing methods such as the Autoregressive Distributed Lag (ARDL), Vector Error Correction Model (VECM), and Dumitrescu-Hurlin (D-H) causality tests, the study provides empirical insights into the interactions among these variables. According to our findings, CO2 emissions decrease when hydropower energy consumption and capital formation rise. In contrast, factors such as fossil fuels, economic growth, agriculture, and labor are associated with rising CO2 emissions. The impulse response analysis indicates that shocks in hydropower energy consumption result in a decrease in CO2 emissions. Additionally, the findings reveal a bidirectional causality between CO2 emissions and hydropower energy consumption. These results underscore the critical importance of sustainable energy policies in SEA countries, promoting investment in hydropower and renewables to mitigate emissions and support environmental sustainability. To meet the region's long-term environmental and economic objectives, it is imperative to reduce reliance on fossil fuels, improve energy efficiency, and promote green technologies. Graphical Abstract


  • Open Access

    Interpretable Machine Learning QSAR Models for Classification and Screening of VEGFR-2 Inhibitors in Anticancer Drug Discovery

    by Teuku Rizky Noviandy, Rinaldi Idroes

    Cancer remains a major global health burden, with angiogenesis playing a central role in tumor growth and progression. Vascular Endothelial Growth Factor Receptor-2 (VEGFR-2) is a key mediator of angiogenesis and an attractive therapeutic target, but existing inhibitors are limited by reduced efficacy, toxicity, and resistance, creating a need for more effective predictive models in drug discovery. In this study, an interpretable machine learning based QSAR approach was developed using a curated dataset of 10,221 VEGFR-2 inhibitors from ChEMBL represented by 164 molecular descriptors. Four algorithms, kNN, AdaBoost, Random Forest, and XGBoost, were compared, and XGBoost achieved the best results with an accuracy of 83.67 percent, sensitivity of 91.38 percent, specificity of 71.73 percent, F1-score of 87.17 percent, and AUC of 0.9009. Model interpretation with LIME identified molecular descriptors related to hydrogen bonding, electrostatics, and lipophilicity as key contributors to activity. These results indicate that interpretable ensemble models can combine strong predictive performance with mechanistic insights, supporting rational design and optimization of novel VEGFR-2 inhibitors for anticancer therapy.


  • Open Access

    Credit Card Fraud Detection Through Explainable Artificial Intelligence for Managerial Oversight

    by Muksalmina Muksalmina, Ahmad Syahyana, Ferdy Hidayatullah, Ghalieb Mutig Idroes, Teuku Rizky Noviandy

    As digital payment systems grow in volume and complexity, credit card fraud continues to be a significant threat to financial institutions. While machine learning (ML) has emerged as a powerful tool for detecting fraudulent activity, its adoption in managerial settings is hindered by a lack of transparency and interpretability. This study examines how explainable artificial intelligence (XAI) can enhance managerial oversight in the deployment of ML based fraud detection systems. Using a publicly available, simulated dataset of credit card transactions, we developed and evaluated four ML models: Logistic Regression, Naïve Bayes, Decision Tree, and Random Forest. Performance was assessed using standard metrics, including accuracy, precision, recall, and F1-score. The Random Forest model demonstrated superior classification performance but also presented significant interpretability challenges due to its complexity. To fill this gap, we applied SHAP (SHapley Additive exPlanations), a leading method for explaining the outputs of the Random Forest model. SHAP analysis revealed that transaction amount and merchant category were the most influential features in determining the risk of fraud. SHAP plots were used to make these insights accessible to non-technical stakeholders. The findings underscore the importance of XAI in promoting transparency, facilitating regulatory compliance, and fostering trust in AI-driven decisions. This study offers practical guidance for managers, auditors, and policymakers seeking to integrate explainable ML tools into financial risk management processes, ensuring that technological advancements are balanced with accountability and informed human oversight.


  • Open Access

    Inductive Biases in Feature Reduction for QSAR: SHAP vs. Autoencoders

    by Teuku Rizky Noviandy, Ghifari Maulana Idroes, Andi Lala, Zuchra Helwani, Rinaldi Idroes

    Machine learning models in drug discovery often depend on high-dimensional molecular descriptors, many of which may be redundant or irrelevant. Reducing these descriptors is essential for improving model performance, interpretability, and computational efficiency. This study compares two widely used reduction strategies: SHAP-based feature selection and autoencoder-based compression, within the context of Quantitative Structure-Activity Relationship (QSAR) classification. LightGBM is used as a consistent modeling framework to evaluate models trained on all descriptors, the top 50 and 100 SHAP-ranked descriptors, and a 64-dimensional autoencoder embedding. The results show that SHAP-based selection produces interpretable and stable models with minimal performance loss, particularly when using the top 100 descriptors. In contrast, the autoencoder achieves the highest test performance by capturing nonlinear patterns in a compact, low-dimensional representation, although this comes at the cost of interpretability and consistency across data splits. These findings reflect the differing inductive biases of each method. SHAP prioritizes sparsity and attribution, while autoencoders focus on reconstruction and continuity. The analysis emphasizes that descriptor reduction strategies are not interchangeable. SHAP-based selection is suitable for applications where interpretability and reliability are essential, such as in hypothesis-driven or regulatory settings. Autoencoders are more appropriate for performance-driven tasks, including virtual screening. The choice of reduction strategy should be guided not only by performance metrics but also by the specific modeling requirements and assumptions relevant to cheminformatics workflows.


  • Open Access

    Explainable Deep Learning with Lightweight CNNs for Tuberculosis Classification

    by Teuku Rizky Noviandy, Ghazi Mauer Idroes, Teuku Zulfikar, Rinaldi Idroes

    Tuberculosis (TB) remains a major global health threat, particularly in low-resource settings where timely diagnosis is critical yet often limited by the lack of radiological expertise. Chest X-rays (CXRs) are widely used for TB screening, but manual interpretation is prone to errors and variability. While deep learning has shown promise in automating CXR analysis, most existing models are computationally intensive and lack interpretability, limiting their deployment in real-world clinical environments. To address this gap, we evaluated three lightweight and explainable CNN architectures, ShuffleNetV2, SqueezeNet 1.1, and MobileNetV3, for binary TB classification using a locally sourced dataset of 3,008 CXR images. Using transfer learning and Grad-CAM for visual explanation, we show that MobileNetV3 and ShuffleNetV2 achieved perfect test performance with 100% accuracy, sensitivity, specificity, precision, and F1-score, along with AUC scores of 1.00 and inference times of 94.66 and 103.63 seconds, respectively. SqueezeNet performed moderately, with a lower F1-score of 82.98% and several misclassifications. These results demonstrate that lightweight CNNs can deliver high diagnostic accuracy and transparency, supporting their use in scalable, AI-assisted TB screening systems for underserved healthcare settings.


  • Open Access

    Predicting AXL Tyrosine Kinase Inhibitor Potency Using Machine Learning with Interpretable Insights for Cancer Drug Discovery

    by Teuku Rizky Noviandy, Ghifari Maulana Idroes, Essy Harnelly, Irma Sari, Fazlin Mohd Fauzi, Rinaldi Idroes

    AXL tyrosine kinase plays a critical role in cancer progression, metastasis, and therapy resistance, making it a promising target for therapeutic intervention. However, traditional drug discovery methods for developing AXL inhibitors are resource-intensive, time-consuming, and often fail to provide detailed insights into molecular determinants of potency. To address this gap, we applied machine learning techniques, including Random Forest, Gradient Boosting, Support Vector Regression, and Decision Tree models, to predict the potency (pIC50) of AXL inhibitors using a dataset of 972 compounds with 550 molecular descriptors. Our results demonstrate that the Random Forest model outperformed others with an R² of 0.703, MAE of 0.553, RMSE of 0.720, and PCC of 0.841, showcasing strong predictive accuracy. SHAP analysis identified critical molecular features, such as RNCG and TopoPSA(NO), as key contributors to inhibitor potency, providing interpretable insights into structure-activity relationships. These findings highlight the potential of machine learning to accelerate the identification and optimization of AXL inhibitors, bridging the gap between computational predictions and rational drug design and paving the way for effective cancer therapeutics.