EduMatika: Jurnal MIPA

Analysis of the Most Dominant Causing Factors of Divorce in 34 Provinces in Indonesia Using the XGBoost Algorithm

Mon, 30 Jun 2025 00:00:00 +0700

Divorce in Indonesia shows a significant increasing trend and has a broad social impact. This study aims to identify the most dominant causes of divorce in 34 provinces in Indonesia using the Extreme Gradient Boosting (XGBoost) machine learning algorithm. Secondary quantitative data from the Central Bureau of Statistics in 2024 were analyzed by pre-processing, data sharing, model training, and performance evaluation. The results showed that constant disputes and quarrels were the main causes of divorce, followed by substance abuse and forced marriage. The developed XGBoost model achieved 75% accuracy in classifying the level of divorce risk. These findings provide new insights into understanding the social factors that influence divorce and can be the basis for designing more effective prevention strategies.

Analysis of Book Production Quality Using Ewma Control Chart: Comparison of Alpha Parameters, Trend Prediction, and Production Correlation

Amelia Putri, Aulia Hani, Anatasia Faradhilla, Risca Octaviyani Hutapea — Mon, 30 Jun 2025 00:00:00 +0700

In the printing industry, maintaining product quality is a critical factor for ensuring customer satisfaction and competitive advantage. This study applies the Exponentially Weighted Moving Average (EWMA) control chart to monitor and control the quality of book production at CV Renjana Offset. Using secondary data from November 2023 to July 2024, the analysis focuses on the percentage of rejected products as a key indicator of quality performance. The EWMA method, with a smoothing constant of ? = 0.3, effectively highlights small shifts in the production process. Results show that all EWMA values lie within the control limits (UCL = 22.54%, LCL = 0.00%), indicating the process is statistically under control. Furthermore, comparison of different alpha values demonstrates the trade-off between sensitivity and stability. A moderate negative correlation (r = -0.505) between production volume and reject percentage suggests increased efficiency at higher production scales. The predicted reject percentage for August 2024 is 4.74%, indicating process stability. Overall, EWMA proves to be a valuable tool in continuous quality monitoring and data-driven decision-making in book production.

The Use of Number Cards to Improve the Beginning Counting Skills of Second Grade Students at SD Muhammadiyah 01 Medan

Indah Setiawani, Elfrianto Elfrianto — Mon, 30 Jun 2025 00:00:00 +0700

This study was initiated by the issue of low arithmetic skills among students in mathematics learning. The objective of this research was to explore the extent to which early arithmetic skills could be improved through the implementation of number card media among second-grade students at SD Muhammadiyah 01 Medan. The method employed was Classroom Action Research (CAR), conducted over two Cycles. Each Cycle consisted of the stages of planning, action implementation, observation, and reflection. Data collection techniques included direct observation and documentation. The findings revealed that the integration of number card media had a significant impact on enhancing students’ early arithmetic skills. This was evidenced by an increase in the percentage of students achieving learning mastery from 47% in Cycle I to 96% in Cycle II. Thus, there was a 49% improvement, indicating that the predetermined success indicators had been successfully met.

Development of Deep Learning Model Based on Convolutional Neural Network (CNN) for Brain Tumor Classification Using MRI Images

Mon, 30 Jun 2025 00:00:00 +0700

Brain tumor classification using MRI images presents a critical challenge in medical radiology. This study develops a deep learning model based on Convolutional Neural Network (CNN) to classify brain MRI images into four categories: Normal, Glioma, Meningioma, and Pituitary. A publicly available dataset from Kaggle consisting of 20,672 images was used, with preprocessing and data augmentation applied. The model architecture includes convolutional, pooling, flatten, dense, and dropout layers, optimized using the Adam optimizer and categorical crossentropy loss function. The evaluation results show that the model achieved an overall accuracy of 96% with high f1-scores across all classes, particularly for the Pituitary class (0.98). The main contribution of this study lies in the integration of diverse data augmentation techniques and Explainable AI (XAI) methods, enabling the visualization of key areas in MRI images that support classification decisions. The proposed model is not only accurate but also demonstrates strong generalization and interpretability, making it a promising tool for clinical decision support systems in brain tumor diagnosis.

Stock Closing Price Prediction of PT Bank Central Asia Tbk (BBCA) with Long Short-Term Memory (LSTM)

Mon, 30 Jun 2025 00:00:00 +0700

Stock price volatility remains one of the key challenges for investors in making accurate investment decisions in Indonesia’s capital market. To address this issue, predictive approaches based on machine learning—such as the Long Short-Term Memory (LSTM) algorithm—are increasingly utilized due to their effectiveness in processing time series data. This study aims to develop a model for predicting the closing price of PT Bank Central Asia Tbk (BBCA) shares using the LSTM method. The dataset consists of historical daily stock prices of BBCA from 2015 to mid-2025, obtained from Yahoo Finance. The research stages include data preprocessing, normalization, sequence generation, LSTM model construction, training and validation, and performance evaluation using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). The results show that the LSTM model successfully predicted closing stock prices with high accuracy, as indicated by a very low validation loss and prediction curves that closely follow actual price trends. This suggests that LSTM has a strong generalization ability and is effective in capturing complex stock movement patterns. The novelty of this research lies in the practical implementation of LSTM for BBCA stock price prediction and its potential application in real-time decision support systems for investors.