Evaluating the effect of feature scaling on L2-regularized logistic regression for breast cancer classification
1 Adekunle Ajasin University, Akungba-Akoko, Ondo State, Nigeria
* Corresponding author: oluwaseun.akinduko@aaua.edu.ng
* Corresponding author: oluwaseun.akinduko@aaua.edu.ng
Abstract
Feature scaling is a common preprocessing technique in machine learning but its effect depends on the scaling strategy and the type of data used. In this paper, a performance test of L2-regularized Logistic Regression was carried out on breast cancer classification, based on three feature-scaling conditions, namely, no scaling, Min--Max scaling, and Z-score standardization. The Wisconsin Diagnostic Breast Cancer (WDBC) dataset, consisting of 569 observations and 30 predictors was used. By using five-fold cross-validation, all the three conditions were examined based on the same model specification, the same folds and the same evaluation criteria. In order to prevent leakage of data, only the training portion of each fold was used to evaluate the scaling parameters, and later applied to its validation part. The performance of the model was evaluated based on accuracy, sensitivity, specificity, F1-score, receiver operating characteristic area under the curve (ROC-AUC), convergence, and computational measures. The findings of this research showed that Z-score standardization recorded the highest mean accuracy of 0.9737, sensitivity of 0.9436, specificity of 0.9916, F1-score of 0.9638, and ROC-AUC of 0.9937. Min--Max scaling recorded the highest mean specificity of 0.9972, the lowest mean sensitivity of 0.8635 and F1-score of 0.9242, and the least computational cost. Across the five folds, all the three scaling conditions achieved a 100\% convergence rate. The findings of this study also revealed that different feature-scaling conditions can perform differently when applied to the same L2-regularized Logistic Regression model, though this is applicable only to the WDBC dataset and the experimental configuration under consideration.
Keywords
Feature scaling
L2 regularization
Logistic Regression
breast cancer classification
Min--Max scaling
Z-score standardization
References
- E. Christodoulou, J. Ma, G. S. Collins, E. W. Steyerberg, J. Y. Verbakel, and B. Van Calster, A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models, J. Clin. Epidemiol., 110 (2019), 12--22. https://doi.org/10.1016/j.jclinepi.2019.02.004
- S. le Cessie and J. C. van Houwelingen, Ridge estimators in logistic regression, J. R. Stat. Soc. Ser. C (Appl. Stat.), 41(1) (1992), 191--201. https://doi.org/10.2307/2347628
- X. H. Cao, I. Stojkovic, and Z. Obradovic, A robust data scaling algorithm to improve classification accuracies in biomedical data, BMC Bioinformatics, 17 (2016), 359. https://doi.org/10.1186/s12859-016-1236-x
- Henderi, T. Wahyuningsih, and E. Rahwanto, Comparison of Min-Max normalization and Z-Score normalization in the K-nearest neighbor (kNN) algorithm to test the accuracy of types of breast cancer, Int. J. Informatics Inf. Syst., 4(1) (2021), 13--20. https://doi.org/10.47738/ijiis.v4i1.73
- A. Adeyemo, H. Wimmer, and L. M. Powell, Effects of normalization techniques on logistic regression in data science, J. Inf. Syst. Appl. Res., 12(2) (2019), 37--44.
- R. Entezari, Breast cancer diagnosis via classification algorithms, arXiv:1807.01334 (2018). https://doi.org/10.48550/arXiv.1807.01334
- W. Cheng and Z. Yu, A fast and interpretable logistic regression framework for breast tumor classification using the Wisconsin Diagnostic Dataset, medRxiv (2025). https://doi.org/10.64898/2025.12.23.25342946
- M. Kumar and V. Bhardwaj, Evaluating label encoding and preprocessing techniques for breast cancer prediction using machine learning algorithms, Int. J. Comput. Intell. Syst., 18 (2025), 218. https://doi.org/10.1007/s44196-025-00957-7
- M. E. Yildirim and Y. B. Salman, Breast cancer classification using z-score thresholding and machine learning, J. Electron. Electromed. Eng. Med. Inform., 7(4) (2025), 1355--1366. https://doi.org/10.35882/jeeemi.v7i4.1165
- J. M. H. Pinheiro, S. V. B. de Oliveira, T. H. S. Silva, P. A. R. Saraiva, E. F. de Souza, R. V. Godoy, L. A. Ambrosio, and M. Becker, The impact of feature scaling in machine learning: Effects on regression and classification tasks, IEEE Access, 13 (2025), 199903--199931. https://doi.org/10.1109/ACCESS.2025.3635541
- R. Watrianthos, Rayendra, E. Asri, Yuhefizar, and Humaira, Threshold-optimized and calibrated logistic regression for breast cancer classification, Salud, Ciencia y Tecnología, 5(2241)(2025). https://doi.org/10.56294/saludcyt20252241
- W. Wolberg, O. Mangasarian, N. Street, and W. Street, Breast Cancer Wisconsin (Diagnostic) [Dataset], UCI Machine Learning Repository (1993). https://doi.org/10.24432/C5DW2B
- W. N. Street, W. H. Wolberg, and O. L. Mangasarian, Nuclear feature extraction for breast tumor diagnosis, in Biomedical Image Processing and Biomedical Visualization, vol. 1905, SPIE, Bellingham, WA, USA (1993), 861--870. https://doi.org/10.1117/12.148698
- T. Fawcett, An introduction to ROC analysis, Pattern Recognit. Lett., 27(8) (2006), 861--874. https://doi.org/10.1016/j.patrec.2005.10.010
How to Cite
Onuoha, O. B. (2026). Evaluating the effect of feature scaling on L2-regularized logistic regression for breast cancer classification. Journal of Physical Sciences: Theories, Methods and Applications, 1(1), 79-87.
O. B. Onuoha, "Evaluating the effect of feature scaling on L2-regularized logistic regression for breast cancer classification," Journal of Physical Sciences: Theories, Methods and Applications, vol. 1, no. 1, pp. 79-87, November 2026.