Implementasi K-Means Clustering dalam Memetakan Karakteristik Performa dan Konsistensi Pembalap pada Era Regulasi Baru Formula 1 2026

Elsa Anggraini (Universitas Bina Sarana Informatika)

JUTEI (Jurnal Terapan Teknologi Informasi)
Vol 10 No 1 (2026): Jurnal Terapan Teknologi Informasi

The 2026 Formula 1 season introduces a radical regulatory transition, rendering historical performance data obsolete. This study addresses the "cold-start" problem in sports analytics by implementing the K-Means clustering algorithm to map competitive hierarchies during the 2026 Bahrain pre-season tests. The analysis is exclusively based on four key performance features: Fastest Lap, Average Lap Time, Standard Deviation (consistency), and Total Laps (reliability), extracted via the FastF1 API. A total of 3,624 telemetry data rows were processed and normalized using StandardScaler. The Elbow Method identified K=4 as the optimal cluster configuration. Although the Silhouette Coefficient of 0.350 reflects the inherent "noise" and "sandbagging" strategies of F1 testing, the model successfully differentiated four distinct performance tiers: Top-Tier Leaders, Stable Midfielders, Reliability-Focused Testers, and Technical Anomalies (Strugglers). The findings provide an objective, data-driven framework for interpreting competitive strength without relying on subjective media reports, proving that unsupervised learning can extract meaningful patterns from unlabelled telemetry data in highly volatile regulatory environments.