Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Stroke Classification Comparison with KNN through Standardization and Normalization Techniques

Rektorat UDINUS 2024 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2024
Goals
Record
79033a98-1cb2-4fff-a71b-f54e060fe57c
Published
2026-10-30

Universitas Dian Nuswantoro conducted a research activity examining the impact of data-scaling techniques on K-Nearest Neighbors (KNN) classification for stroke data. The activity involved UDINUS researchers Firmansyah, Muhammad Raihan, and Astuti, Yani Parti as participants. It focused on managing diverse scales among health attributes in a stroke dataset, with the stated aim of improving the accuracy and reliability of the classification model.

The activity compared three preprocessing conditions: z-score standardization, min-max normalization, and the use of data without scaling. Following these preprocessing conditions, the KNN model was trained and evaluated using various data split ratios. The comparison addressed how alternative approaches to preparing health-related attributes affected the performance of the same classification method. The source describes the activity as a study of stroke classification and data scaling, without providing further details about the dataset composition or the specific evaluation configuration.

The reported results showed comparable performance between z-score standardization and min-max normalization, although slight variations occurred across the data split ratios. Both z-score standardization and min-max normalization achieved an accuracy of 95.07% in the reported results. The source also states that normalization produced a higher average accuracy than standardization, with average accuracies of 94.25% and 94.21%, respectively. These findings identify data scaling as an important consideration in the application of KNN to stroke classification and associate the preprocessing comparison with robust machine-learning performance and informed health decisions.

In relation to the stated goal concerning health, this UDINUS activity addresses a health research focus by examining a computational approach for stroke classification. The activity provides a documented comparison of preprocessing methods and their reported classification performance. Institutionally, it represents research by UDINUS participants addressing a health-related analytical problem through evaluation of KNN and data-scaling techniques.

Reference https://www.scopus.com/pages/publications/85199669018?origin=resultslist