This study aims to describe the implementation of the Fartlek method based on Deep Learning and to improve the basic locomotor movement skills of students in Class 5B at SDN 2 Loktabat Selatan. The study was motivated by the low locomotor movement ability of students, where only 7 out of 27 students (25.93%) achieved mastery in the pre-cycle stage with a class average of 61.11. This research employed a Classroom Action Research (CAR) design using the Kemmis and McTaggart model, conducted over two cycles. The research subjects were 27 students of Class 5B for the 2025/2026 academic year. Sampling was conducted through purposive sampling based on the results of an initial diagnostic observation. Data collection techniques included student activity observation and locomotor skills performance tests. Data were analyzed using comparative descriptive techniques. Results showed significant improvement: classical mastery increased from 25.93% (pre-cycle) to 55.56% (Cycle I) and 81.48% (Cycle II), with average scores improving from 61.11 to 71.60 and 78.09. Student activity improved from Adequate to Very Good (81.48%) at the end of Cycle II. It is concluded that the Fartlek method based on Deep Learning is effective in improving basic locomotor movement skills in elementary school students.
Copyrights © 2026