<![CDATA[This study investigates the difference in Thermodynamics learning outcomes between students taught using a Project-Based Learning (PBL) model integrated with deep learning and those taught using the PBL model alone.. The research employed an experimental method with a quasi-experimental design, utilizing a Nonequivalent Pretest-Posttest Control Group Design. The study was conducted in the Physics Department of Manado State University for Physics Education and Physics programs during the Odd Semester of the 2025/2026 Academic Year. The research population included all active students from these programs, with a random sample of 40 students divided into an experimental class and a control class, each consisting of 20 students. Data were collected through essay-type tests, administered as pretests and posttests, and subsequently analyzed statistically using descriptive and inferential techniques with Python programming. The findings indicate that the experimental class, which implemented deep learning integrated with the Project Based Learning model, achieved a higher average Thermodynamics learning outcome (80.28) compared to the control class (72.35), demonstrating better data consistency (standard deviation of 7.93 versus 9.71). Shapiro-Wilk normality tests for both classes confirmed a normal distribution of data (p-value for experimental class is 0.3147 while control class is 0.0638), and Levene's homogeneity test confirmed homogeneous variances (p-value 0.2529). Furthermore, the independent sample t-test results showed a t-statistic of 2.8289 and a p-value of 0.0074, which is less than 0.05. This leads to the conclusion that there is a statistically significant difference in Thermodynamics learning outcomes between the experimental and control classes. These findings suggest that the integration of deep learning with the Project Based Learning model is effective in enhancing Thermodynamics learning outcomes.]]>
Copyrights © 2025
