Evaluasi Fitur Berbasis Peringkat: Analisis Komparatif SHAP vs 'Gain' LightGBM Menggunakan Evaluasi Iteratif untuk Prediksi Gagal Bayar
DOI:
https://doi.org/10.26418/jp.v11i3.101307Keywords:
prediksi gagal bayar, feature ranking, lightgbm, shap, ketidakseimbangan kelasAbstract
Prediksi gagal bayar dalam bisnis P2P lending sangat penting sebagai bagian manajemen risiko, namun pada prakteknya sering dihadapkan pada tantangan data berdimensi tinggi, ketidakseimbangan kelas, dan kebutuhan interpretasi model black box. Salah satu tantangan utama adalah dilema pemilihan fitur, apakah menggunakan metode Explainable AI (XAI) seperti SHAP yang akurat secara teoritis namun membutuhkan resource komputasi lebih, atau menggunakan metode gain bawaan model seperti LightGBM yang jauh lebih efisien. Penelitian ini bertujuan untuk menentukan secara empiris apakah resource komputasi SHAP memberikan keunggulan kinerja yang signifikan dibandingkan metode gain LightGBM untuk tugas global feature ranking dalam prediksi gagal bayar. Penelitian menggunakan dataset Lending Club, metodologi penelitian membandingkan peringkat fitur dari kedua metode melalui skema forward feature selection iteratif. Pendekatan ini diuji pada data asli dan data yang diseimbangkan menggunakan teknik oversampling ADASYN. Hasil penelitian menunjukkan dua temuan utama. Pertama, ADASYN efektif meningkatkan recall (metrik utama) sekitar 10%. Kedua, kinerja prediktif model yang dihasilkan dari peringkat fitur berbasis gain LightGBM dan berbasis SHAP adalah identik secara statistik, dengan deviasi metrik di bawah 0.6%. Kinerja optimal juga tercapai hanya dengan 15-20 fitur teratas. Kesimpulannya, resource komputasi SHAP tidak dapat dibenarkan untuk tugas global feature ranking dalam skenario ini. Metode gain bawaan LightGBM terbukti memadai dan efisien, berkontribusi pada pengembangan model yang lebih sederhana tanpa mengorbankan daya prediksiReferences
S. Septiani, S. F. Lalita, and D. R. Zahra, “Pengaruh Financial Technology Peer to Peer Lending dan Uang Elektronik (E-Money) terhadap Pertumbuhan Ekonomi di Indonesia (Tahun 2021-2023),” Jurnal Keuangan dan Perbankan, vol. 21, no. 1, pp. 69–78, Dec. 2024, doi: 10.35384/jkp.v21i1.591.
N. Stanley and C. Kohardinata, “Financial Technology Peer to Peer dan Kredit Perbankan: Kompetisi atau Kolaborasi?,” Reviu Akuntansi dan Bisnis Indonesia, vol. 9, no. 1, pp. 40–60, Feb. 2025, doi: 10.18196/rabin.v9i1.23918.
H. Ighni et al., “Peran Ojk Dalam Menangani Kredit Macet Untuk Stabilitas Keuangan Perusahaan: Studi Kasus Pt Tab,” Jurnal Ilmiah Ekonomi Manajemen Bisnis dan Akuntansi, vol. 2, no. 4, pp. 340–350, Jul. 2025, doi: 10.61722/jemba.v2i5.1197.
C. Yuanqing, Z. Gao, Z. Jian, R. Alizadehsani, and P. Pławiak, “Credit risk prediction model for listed companies based on improved reinforcement learning and Bayesian optimization hyperband,” PLoS One, vol. 20, no. 10, p. e0332150, Oct. 2025, doi: 10.1371/journal.pone.0332150.
L. Hoanh-Su, L. Q. C. Phong, T. C. Vinh, H. M. M. Nhat, and J.-H. Lee, “Default Prediction in the Finance Industry Based on Ensemble Learning: Combining Machine Learning and Deep Learning,” Business Systems Research Journal, vol. 16, no. 1, pp. 198–218, Jun. 2025, doi: 10.2478/bsrj-2025-0010.
A. Nigmonov, S. Shams, and K. Alam, “Macroeconomic determinants of loan defaults: Evidence from the U.S. peer-to-peer lending market,” Res Int Bus Finance, vol. 59, p. 101516, Jan. 2022, doi: 10.1016/j.ribaf.2021.101516.
S. M. Fati, “A LOAN DEFAULT PREDICTION MODEL USING MACHINE LEARNING AND FEATURE ENGINEERING,” ICIC Express Letters, vol. 18, no. 1, pp. 27–37, Jan. 2024, doi: 10.24507/icicel.18.01.27.
X.-F. Li, C. Zhang, X.-C. Lin, T.-J. Lv, and L.-L. Liu, “Research on Default Risk of Peer-To-Peer Online Lending Based on Data Mining Algorithm,” J Comput (Taipei), vol. 31, no. 2, pp. 83–100, 2020, doi: 10.3966/199115992020043102009.
A. Akinjole, O. Shobayo, J. Popoola, O. Okoyeigbo, and B. Ogunleye, “Ensemble-Based Machine Learning Algorithm for Loan Default Risk Prediction,” Mathematics, vol. 12, no. 21, Nov. 2024, doi: 10.3390/math12213423.
V. S. F. Gonçalves and V. R. De Carvalho, “A Review of Interpretability Methods for Gradient Boosting Decision Trees,” Journal of the Brazilian Computer Society, vol. 31, no. 1, pp. 640–654, Aug. 2025, doi: 10.5753/jbcs.2025.5324.
Andrew Nii Anang, Oluwatosin Esther Ajewumi, Tobi Sonubi, Kenneth Chukwujekwu Nwafor, John Babatope Arogundade, and Itiade James Akinbi, “Explainable AI in financial technologies: Balancing innovation with regulatory compliance,” International Journal of Science and Research Archive, vol. 13, no. 1, pp. 1793–1806, Oct. 2024, doi: 10.30574/ijsra.2024.13.1.1870.
N. Robinson and N. Sindhwani, “Loan Default Prediction Using Machine Learning,” in 2024 11th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) (ICRITO), IEEE, Mar. 2024, pp. 1–5. doi: 10.1109/ICRITO61523.2024.10522232.
L. Zhu, Y. Bian, and L. You, “FairSHAP: Preprocessing for Fairness Through Attribution-Based Data Augmentation,” Oct. 2025, [Online]. Available: http://arxiv.org/abs/2505.11111
A. V. Konstantinov and L. V. Utkin, “Interpretable machine learning with an ensemble of gradient boosting machines,” Knowl Based Syst, vol. 222, p. 106993, Jun. 2021, doi: 10.1016/j.knosys.2021.106993.
L.-H. Li, A. K. Sharma, and S.-T. Cheng, “Explainable AI based LightGBM prediction model to predict default borrower in social lending platform,” Intelligent Systems with Applications, vol. 26, p. 200514, Jun. 2025, doi: 10.1016/j.iswa.2025.200514.
N. Nguyen and D. Ngo, “Comparative analysis of boosting algorithms for predicting personal default,” Cogent Economics & Finance, vol. 13, no. 1, Dec. 2025, doi: 10.1080/23322039.2025.2465971.
U. Ahmed, A. Mahmood, M. A. Tunio, G. Hafeez, A. R. Khan, and S. Razzaq, “Investigating boosting techniques’ efficacy in feature selection: A comparative analysis,” Energy Reports, vol. 11, pp. 3521–3532, Jun. 2024, doi: 10.1016/j.egyr.2024.03.020.
X. Zhao, Y. Liu, and Q. Zhao, “Improved LightGBM for Extremely Imbalanced Data and Application to Credit Card Fraud Detection,” IEEE Access, vol. 12, pp. 159316–159335, 2024, doi: 10.1109/ACCESS.2024.3487212.
N. Nguyen and D. Ngo, “Comparative analysis of boosting algorithms for predicting personal default,” Cogent Economics & Finance, vol. 13, no. 1, Dec. 2025, doi: 10.1080/23322039.2025.2465971.
M. A. Muslim, Y. Dasril, M. Sam’an, and Y. N. Ifriza, “An improved light gradient boosting machine algorithm based on swarm algorithms for predicting loan default of peer-to-peer lending,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 28, no. 2, p. 1002, Nov. 2022, doi: 10.11591/ijeecs.v28.i2.pp1002-1011.
S. Alfadia Shauqie et al., “Analisis Pengaruh Fitur Terhadap Tinggi Badan Anak menggunakan SHAP,” JEPIN (Jurnal Edukasi dan Penelitian Informatika), vol. 11, no. 2, pp. 271–276, Aug. 2025, doi: https://doi.org/10.26418/jp.v11i2.96734.
R. Mitchell, E. Frank, and G. Holmes, “GPUTreeShap: massively parallel exact calculation of SHAP scores for tree ensembles,” PeerJ Comput Sci, vol. 8, p. e880, Apr. 2022, doi: 10.7717/peerj-cs.880.
A. V. Ponce‐Bobadilla, V. Schmitt, C. S. Maier, S. Mensing, and S. Stodtmann, “Practical guide to SHAP analysis: Explaining supervised machine learning model predictions in drug development,” Clin Transl Sci, vol. 17, no. 11, Nov. 2024, doi: 10.1111/cts.70056.
K. Aas, M. Jullum, and A. Løland, “Explaining individual predictions when features are dependent: More accurate approximations to Shapley values,” Artif Intell, vol. 298, p. 103502, Sep. 2021, doi: 10.1016/j.artint.2021.103502.
A. H. Putra and A. Salam, “A Comparative Performance of SMOTE, ADASYN and Random Oversampling in Machine Learning Models on Prostate Cancer Dataset,” Journal of Applied Informatics and Computing, vol. 9, no. 3, pp. 603–610, Jun. 2025, doi: 10.30871/jaic.v9i3.9308.
F. A. Wenando et al., “Optimizing Hate Speech Detection in Indonesian Social Media: An ADASYN and LSTM-Based Approach,” Journal Européen des Systèmes Automatisés, vol. 58, no. 1, pp. 13–20, Jan. 2025, doi: 10.18280/jesa.580102.
M. Yusuf, A. Haq, and S. Rochimah, “Integrating Adaptive Sampling with Ensembles Model for Software Defect Prediction,” Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control, May 2025, doi: 10.22219/kinetik.v10i2.2191.
A. A. Mustafa, H. M. Hussein, M. M. Kadhim, and M. J. Hussein, “A Hybrid Oversampling Approach for Fraud Detection: Integrating SMOTE-ENN and ADASYN,” International Journal of Safety and Security Engineering, vol. 15, no. 06, pp. 1243–1250, Jun. 2025, doi: 10.18280/ijsse.150614.
Z. Liu, Z. Zhang, H. Yang, G. Wang, and Z. Xu, “An innovative model fusion algorithm to improve the recall rate of peer-to-peer lending default customers,” Intelligent Systems with Applications, vol. 20, p. 200272, Nov. 2023, doi: 10.1016/j.iswa.2023.200272.
Q. G. Nguyen et al., “Enhancing Credit Risk Management with Machine Learning: A Comparative Study of Predictive Models for Credit Default Prediction,” The American Journal of Applied Sciences, vol. 07, no. 01, pp. 21–30, Jan. 2025, doi: 10.37547/tajas/Volume07Issue01-04.
