Implementation of Deep Learning through the Problem-Based Learning Model to Improve Fifth Grade Students’ Science Learning Outcomes

Abstract

Students' learning outcomes in science at the elementary school level were still not optimal, indicating the need for effective learning approaches to improve students' achievement. One approach that can be implemented is deep learning through the Problem Based Learning (PBL) model. This study aimed to improve fifth-grade elementary students’ science learning outcomes through the implementation of deep learning using the Problem-Based Learning model. The study employed Classroom Action Research involving 21 fifth-grade students at SDN Bantarkemang 6. The research was conducted in two cycles from March to April 2026. Data were collected through learning achievement tests and observation sheets, then analyzed using descriptive statistics and Criterion-Referenced Evaluation (PAP). The success criteria were determined based on improvements in students' learning outcomes and the achievement of the school's minimum mastery criterion. The results showed that the percentage of students achieving learning mastery increased from 73.8% in Cycle I (moderately high) to 83.85% in Cycle II (high), representing an improvement of approximately 10%. These findings indicate that implementing deep learning through the Problem-Based Learning model effectively improved fifth-grade elementary students’ science learning outcomes. Practically, by promoting active participation, critical thinking, and meaningful learning experiences, the combined model provides teachers with an effective instructional framework to foster meaningful science comprehension in primary education.