EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Analysis of the increase in model forecasting accuracy after data normalization

Vladyslav Pylypenko, Vladyslava Skidan, Antonina Volivach

Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska · 30.06.2026 · 16 (2) · 102-106

DOI: 10.35784/iapgos.8053 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska, online erschienen am 30.06.2026. Verfasst von Vladyslav Pylypenko und 2 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In modern machine learning, data preprocessing is a crucial step that significantly affects the performance of classification models and the accuracy of prediction. This research investigates the impact of data normalization on the prediction accuracy of two common machine learning algorithms: logistic regression and support vector machine (SVM). Data normalization methods, including logarithmic transformation and scaling methods (such as RobustScaler and MinMaxScaler), were applied to evaluate their impact on model accuracy, balanced accuracy, F1 score, and area under the receiver operating characteristic (AUC). The results showed that the normalization process led to a significant improvement in model performance. In particular, logistic regression showed a moderate increase in all key metrics: accuracy improved by 14.6%, balanced accuracy by 12.3%, F1 score by 11.5%, and AUC by 3.9%. In contrast, SVM showed significant improvements: accuracy increased by 42.5%, balanced accuracy by 81.3%, F1 score by 21.3%, and AUC by 117.9%. These results highlight the importance of preprocessing steps for models sensitive to feature scaling, such as SVM. The ROC curves confirmed the improvement in classification performance, with the AUC of SVM increasing from 0.407 to 0.887 after normalization, indicating a shift from a low-performing model to one that more accurately discriminates between positive and negative classes. This improvement demonstrates the effectiveness of normalization in improving the generalization and robustness of the model, especially when handling unbalanced datasets. The result highlights that data normalization is an important step before training and learning a model. It not only improves the accuracy of the model, but also provides better adaptability and stability in real-world applications. And it is also an important step towards achieving reliable and accurate predictions. The results of the study highlight the importance of data preprocessing in machine learning processes, especially for algorithms that depend on feature scale or geometric methods, such as SVM and logistic regression.

Zitieren

APA

Pylypenko, V., Skidan, V., & Volivach, A. (2026). Analysis of the increase in model forecasting accuracy after data normalization. Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska, 16(2), 102-106. https://doi.org/10.35784/iapgos.8053

Vancouver

Pylypenko V, Skidan V, Volivach A. Analysis of the increase in model forecasting accuracy after data normalization. Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska. 2026;16(2):102-106. doi:10.35784/iapgos.8053

BibTeX
@article{Pylypenko2026,
  author = {Pylypenko, Vladyslav and Skidan, Vladyslava and Volivach, Antonina},
  title = {Analysis of the increase in model forecasting accuracy after data normalization},
  journal = {Informatyka, Automatyka, Pomiary w Gospodarce i Ochronie Środowiska},
  year = {2026},
  volume = {16},
  number = {2},
  pages = {102--106},
  doi = {10.35784/iapgos.8053}
}
Kostenlos starten →