Integrating Deep Learning for Analyzing the Impact of Discussion Method on Student Learning Outcomes in Social Studies

Abstract

This study investigates the impact of the discussion method on students’ learning outcomes in Social Studies by integrating traditional statistical analysis with Deep Learning approaches. The research was conducted at SMP Negeri 5 Pontianak with a total of 141 seventh-grade students as participants, selected through total sampling. Using a quantitative ex post facto design, data were collected through questionnaires measuring students’ engagement in discussions and documentation of Social Studies test scores. Analysis involved simple linear regression to determine the influence of discussion on achievement, complemented by a Deep Neural Network model to enhance predictive accuracy. The results indicate that the discussion method significantly improves student performance, with most students achieving scores in the “good” and “very good” categories. Furthermore, the Deep Learning model demonstrated high predictive accuracy, highlighting that students’ ability to respond to questions and problem-solving skills were the most influential indicators of achievement. These findings confirm the effectiveness of interactive pedagogy in fostering academic success while offering new insights into how artificial intelligence can enrich educational research. The study contributes to both theory and practice by demonstrating the potential of combining traditional classroom strategies with modern analytics to support evidence-based teaching and adaptive learning.