Skripsi

OPTIMASI NILAI K PADA KNN MENGGUNAKAN PARTICLE SWARM OPTIMIZATION UNTUK DIAGNOSIS STROKE

---

Stroke is a leading cause of disability and death worldwide, making accurate early diagnosis essential to reduce long-term impacts. This study aims to implement K-Nearest Neighbors (KNN) for Stroke diagnosis Classification and to optimize the value of K using Particle Swarm Optimization (PSO). The dataset was obtained from Kaggle, consisting of 5,110 entries and 12 demographic and clinical features. The preprocessing stages included handling missing values, standardization, categorical feature encoding, and outlier adjustment using z-score, while SMOTE was applied to balance class distribution. The experimental results of KNN show the highest accuracy of 87.9% and an F1-score of 0.896 at K=1 with a 90:10 ratio, although performance decreased as the value of K increased. In contrast, optimizing the value of K with PSO allowed a more adaptive and stable parameter selection, as the algorithm explored the best configuration through a combination of population size, inertia weight, cognitive coefficient, social coefficient, and number of iterations. At the 70:30 ratio, the optimal configuration with population = 20, W = 0.1, C1 = 3, C2 = 1, and 30 iterations achieved an accuracy of 83.43% and an F1-score of 0.857. The integration of KNN and PSO not only improved accuracy but also produced a more consistent model across different data distribution ratios. Therefore, this study demonstrates that the combination of KNN and PSO has the potential to be an effective solution for supporting Stroke diagnosis based on medical data, and it can be further developed in disease prediction systems using machine learning.

---

Availability

Detail Information

Series Title

-

Call Number

T1889862025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xiii, 142 hlm.; ilus.; tab, 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.310 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Informatika
Pembelajaran Mesin

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment