K-Nearest Neighbor Algorithm Outperforms Decision Tree in Data Mining-Based Early Detection of Breast Cancer

Illustration by AI

FORMOSA NEWS

Recent research conducted by Asri Liya Astuti from Universitas Pelita Bangsa evaluates the effectiveness of two artificial intelligence algorithms, namely K-Nearest Neighbor (KNN) and Decision Tree C4.5, in classifying breast cancer cases. Published in July 2026, this research is crucial for improving the accuracy of early diagnosis to reduce mortality rates from one of the most dangerous diseases for women.

Breast cancer ranks at the top as the leading cause of cancer-related death among women in Indonesia, with an incidence rate reaching 42.1 per 100,000 population. Many patients often overlook the potential recurrence of this disease because small remaining amounts of cancer cells after treatment can still trigger serious health problems. Therefore, the utilization of data analytics technology is essential to help recognize tumor patterns quickly and objectively.

In its analytical process, this research applies the Cross-Industry Standard Process for Data Mining (CRISP-DM) industrial standard framework with the assistance of RapidMiner software. The data used originates from the Breast Cancer Wisconsin dataset, which includes 699 medical record entries to categorize tumor types into benign and malignant groups.

The performance comparison results reveal highly significant findings from both computational models:

• The K-Nearest Neighbor (KNN) algorithm proves to be the best model with an accuracy rate reaching 97.14%.
• The KNN model also records an Area Under Curve (AUC) score of 0.976, which is classified in the "excellent" classification tier.
• Meanwhile, the Decision Tree C4.5 algorithm records an accuracy rate of 95.71% with an AUC score of 0.957.

According to Asri Liya Astuti from Universitas Pelita Bangsa, this performance advantage proves that the KNN approach is a reliable and practical instrument to accurately differentiate breast cancer cases. These findings are expected to strengthen technology implementation in the health sector, help medical personnel formulate more precise clinical decisions, and enhance the effectiveness of early detection programs in society.

Author Profiles

• Full Name: Asri Liya Astuti
• University Affiliation: Universitas Pelita Bangsa, Bekasi Regency

Research Sources

• Article Title: Classification Analysis of 2 K-Nearest Neighbor (KNN) and Decision Tree Algorithms Using Rapidminer in Breast Cancer
• Journal Name: International Journal of Education and Life Sciences (IJELS), Vol. 4, No. 7
• Publication Year: July 2026
• DOI: https://doi.org/10.59890/ijels.v4i7.44
• URL: https://ntlmultitechpublisher.my.id/index.php/ijels