Enhancing Loan Approval Decision-Making: An Interpretable Machine Learning Approach Using Lightgbm for Digital Economy Development

,; Noviandy, Teuku Rizky; Idroes, Ghalieb Mutig; ,; Hardi, Irsan; ,

doi:10.24191/mjoc.v9i1.25691

Cited by 7 publications

(5 citation statements)

References 30 publications

(31 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We employed a grid search approach to systematically evaluate the combination of hyperparameters. This method involves exhaustively searching through a manually specified subset of the hyperparameter space [23]. The hyperparameters tuned and their respective ranges are summarized in Table 2.…”

Section: Hyperparameter Tuningmentioning

confidence: 99%

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Safhadi,

Noviandy,

Irvanizam

et al. 2024

Infolitika J. Data Sci.

Self Cite

View full text Add to dashboard Cite

The identification of chemical compounds in essential oils is crucial in industries such as pharmaceuticals, perfumery, and food. Kovats Retention Index (RI) values are essential for compound identification using gas chromatography-mass spectrometry (GC-MS). Traditional RI determination methods are time-consuming, labor-intensive, and susceptible to experimental variability. Recent advancements in data science suggest that artificial intelligence (AI) can enhance RI prediction accuracy and efficiency. However, the full potential of AI, particularly artificial neural networks (ANN), in predicting RI values remains underexplored. This study develops a backpropagation neural network (BPNN) model to predict the Kovats RI values of essential oil compounds using five molecular descriptors: ATSc1, VCH-7, SP-1, Kier1, and MLogP. We trained the BPNN on a dataset of 340 essential oil compounds and optimized it through hyperparameter tuning. We show that the optimized BPNN model, with an epoch count of 100, a learning rate of 0.1, a hidden layer size of 10 neurons, and the ReLU activation function, achieves an R² value of 0.934 and a Root Mean Squared Error (RMSE) of 76.98. These results indicate a high correlation between predicted and actual RI values and a low average prediction error. Our findings demonstrate that BPNNs can significantly improve the efficiency and accuracy of compound identification, reducing reliance on traditional experimental methods.

show abstract

Section: Hyperparameter Tuningmentioning

confidence: 99%

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Safhadi,

Noviandy,

Irvanizam

et al. 2024

Infolitika J. Data Sci.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Meanwhile, the testing set was reserved for evaluating the trained model's performance on unseen data, providing an unbiased estimate of its generalization ability. This separation ensures the model's performance can be accurately assessed and helps prevent overfitting the training data [28].…”

Section: Datasetmentioning

confidence: 99%

Machine Learning for Early Detection of Dropout Risks and Academic Excellence: A Stacked Classifier Approach

Noviandy,

Zahriah,

Yandri

et al. 2024

J. Educ. Manag. Learn.

Self Cite

View full text Add to dashboard Cite

Education is important for societal advancement and individual empowerment, providing opportunities, developing essential skills, and breaking cycles of poverty. Nonetheless, the path to educational success is marred by challenges such as achieving academic excellence and preventing student dropouts. Early identification of students at risk of dropping out or those likely to excel academically can significantly enhance educational outcomes through tailored interventions. Traditional methods often fall short in precision and foresight for effective early detection. While previous studies have utilized machine learning to predict student performance, the potential for more sophisticated ensemble methods, such as stacked classifiers, remains largely untapped in educational contexts. This study develops a stacked classifier integrating the predictive strengths of LightGBM, Random Forest, and logistic regression. The model achieved an accuracy of 80.23%, with precision, recall, and F1-score of 79.09%, 80.23%, and 79.20%, respectively, surpassing the performance of the individual models tested. These results underscore the stacked classifier's enhanced predictive capability and transformative potential in educational settings. By accurately identifying students at risk and those likely to achieve academic excellence early, educational institutions can better allocate resources and design targeted interventions. This approach optimizes educational outcomes and supports informed policymaking, fostering environments conducive to student success.

show abstract

“…Standardization ensures that all features have comparable scales, enhancing ML algorithms' stability and convergence. By executing these preprocessing steps, we improve the quality and suitability of the dataset for subsequent model training and evaluation [28].…”

Section: Data Preprocessingmentioning

confidence: 99%

Interpretable Machine Learning for Chronic Kidney Disease Diagnosis: A Gaussian Processes Approach

Noviandy,

Idroes,

Syukri

et al. 2024

Indo. Jour. of Case Rep.

Self Cite

View full text Add to dashboard Cite

Chronic Kidney Disease (CKD) is a global health issue impacting over 800 million people, characterized by a gradual loss of kidney function leading to severe complications. Traditional diagnostic methods, relying on laboratory tests and clinical assessments, have limitations in sensitivity and are prone to human error, particularly in the early stages of CKD. Recent advances in machine learning (ML) offer promising tools for disease diagnosis, but a lack of interpretability often hinders their adoption in clinical practice. Gaussian Processes (GP) provide a flexible ML model capable of delivering predictions and uncertainty estimates, essential for high-stakes medical applications. However, the integration of GP with interpretable methods remains underexplored. We developed an interpretable CKD classification model to address this knowledge gap by combining GP with Shapley Additive Explanations (SHAP). We assessed the model's performance using three GP kernels (Radial Basis Function, Matern, and Rational Quadratic). The results show that the Rational Quadratic kernel outperforms the other kernels, achieving an accuracy of 98.75%, precision of 100%, sensitivity of 97.87%, specificity of 100%, and an F1-score of 98.51%. SHAP values indicate that haemoglobin and specific gravity are the most influential features. The results demonstrate that the Rational Quadratic kernel enhances predictive accuracy and provides robust uncertainty estimates and interpretable explanations. This combination of accuracy and interpretability supports clinicians in making informed decisions and improving patient management and outcomes in CKD. Our study connects advanced ML techniques with practical medical applications, leading to more effective and reliable ML-driven healthcare solutions.

show abstract

Enhancing Loan Approval Decision-Making: An Interpretable Machine Learning Approach Using Lightgbm for Digital Economy Development

Cited by 7 publications

References 30 publications

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Machine Learning for Early Detection of Dropout Risks and Academic Excellence: A Stacked Classifier Approach

Interpretable Machine Learning for Chronic Kidney Disease Diagnosis: A Gaussian Processes Approach

Contact Info

Product

Resources

About