<![CDATA[Problem-solving skills (PSS) are one of the key competencies students need to meet the demands of the 21st century; however, field observations indicate that students’ PSS remain low, particularly in thermodynamics. This study aims to improve students’ PSS through the application of the Problem-Based Learning (PBL) model with a Deep Learning approach. The method used is quantitative with a quasi-experimental design, specifically a Non-Equivalent Control Group Design, involving 59 11th-grade science students at a public high school in Subang Regency, divided into an experimental class (PBL with Deep Learning) and a control class (PBL only). The instrument consisted of 22 PBL essay test questions based on five aspects according to Heller, namely focus on the problem, describe the physics, plan a solution, execute the plan, and evaluate the answer. The results showed that both classes experienced an increase in PBL skills, with an N-Gain of 0.47 for the experimental class and 0.35 for the control class (both in the moderate category). The Mann-Whitney U test showed an Asymp. Sig (2-tailed) < 0.001, indicating a significant difference between the two classes. These results prove that PBM with a Deep Learning approach is more effective in addressing students’ low KPM in thermodynamics material compared to PBM alone.]]>
Copyrights © 2026
