Jurnal Aplikasi Sains Data

Application of Convolutional Neural Network (CNN) for Web-Based Translation of Indonesian Text into Sign Language

Diajeng Prameswari — 2025-10-28

Communication for the deaf and hard of hearing is often hindered by the limited number of sign language interpreters. This research aims to develop a web-based text-to-text sign language translation system using Convolutional Neural Networks (CNN) to bridge this communication gap. The system is built with the ASL Alphabet dataset containing 87,000 images from 29 classes (A-Z, SPACE, DELETE, NOTHING). The CNN model was designed with three convolutional layers and trained for 15 epochs using 80% of the data, while 20% of the data was used for testing. The user interface was developed using Streamlit for ease of use. Training results showed a training accuracy of 98.96% and a validation accuracy of 98.61% at the 15th epoch. Model evaluation yielded an overall accuracy of 98%, with high precision, recall, and F1-score values for most classes. This research demonstrates the significant potential of CNN in developing automatic sign language translators, which is expected to improve information accessibility and inclusivity for the deaf community.

Statistical Analysis of Infant Malnutrition Cases in North Sumatra Before and After COVID-19 Using the Wilcoxon Test

Desi Daomara Sitanggang — 2025-10-28

Child malnutrition remains a very important public health issue in Indonesia. Malnutrition is a condition of deficiency in energy and essential nutrients that can lead to impaired physical growth, mental development, and an increased risk of mortality in children. The prevalence of malnutrition among toddlers in Indonesia is still quite high and shows disparities between regions, especially in provinces with high poverty rates. One province of concern is North Sumatra, which, according to data from the Ministry of Health, has had a significant incidence of malnutrition in the last five years. This condition was exacerbated by the emergence of the COVID-19 pandemic at the end of 2019, which has had a major impact on various sectors of life, including family health and economy. The pandemic caused significant disruptions to primary healthcare systems, including a decrease in posyandu activities, immunizations, and monitoring of children's nutritional status. The decline in household income during the pandemic made it difficult for families to meet their balanced nutritional food needs. A UNICEF study showed an increased risk of acute malnutrition in children during the pandemic, especially in previously vulnerable areas. To measure the impact of the COVID-19 pandemic on the incidence of child malnutrition, a statistical approach that can compare data before and after the pandemic is needed. This study aims to analyze the difference in the incidence of child malnutrition before and after the COVID-19 pandemic in North Sumatra Province using the Wilcoxon test method. Using the Wilcoxon Signed-rank Test statistical method, a comparative analysis was performed between the medians of the data from 2018 and 2023. The results of the study showed that there was a difference between the medians of the two data sets.

Feature Importance-Guided Ensemble Classification for Predicting Recurrence in Differentiated Thyroid Cancer

Muhammad Ghinan Navsih — 2025-10-28

Accurate prediction of cancer recurrence is critical for improving patient monitoring and personalized treatment planning. In this study, we propose a machine learning framework to predict recurrence in patients with differentiated thyroid cancer using statistically selected clinical features. Feature relevance was assessed using ANOVA for ordinal/numerical variables and the Chi-square test for one-hot encoded categorical variables, allowing us to identify the most informative predictors. We then trained three distinct classifiers—Random Forest, Logistic Regression, and XGBoost—and combined them using a hard voting ensemble strategy. The proposed ensemble achieved an accuracy of 98.7% on the test set, with particularly strong precision and recall scores for the recurrent class, indicating its potential clinical utility. Interestingly, all three base classifiers produced identical predictions on the test data, suggesting the dataset’s strong internal structure and the effectiveness of our feature selection process. This work highlights the value of integrating statistical feature selection with ensemble modeling for robust and interpretable prediction in clinical oncology applications.

Application of XGBoost for Risk Level Classification of Fires in Surabaya City in 2024 and Interactive Spatial Visualization Based on Streamlit

Sarah Aprilia Hasibuan Sarah — 2025-10-28

Fire in urban areas such as Surabaya City is a non-natural disaster that can have a
significant impact on public safety, economic stability, and the environment. This study aims to
develop a fire risk level classification model using Extreme Gradient Boosting (XGBoost) algorithm
based on selected predictor variables, namely response time, fire subtype, and number of victims
affected. The dataset consists of 859 fire events throughout 2024, enriched with spatial and
demographic attributes. The research methodology involved data preprocessing (including label coding
and normalization), class imbalance handling with Synthetic Minority Over-sampling Technique
(SMOTE), model training with XGBoost, and evaluation using metrics such as accuracy, precision,
recall, and f1-score. The classification model achieved excellent performance, with an overall accuracy
of 1.00% and perfect precision, recall, and f1-score of 1.00 across all risk categories (low, medium,
and high). Confusion matrix and ROC curve analysis confirmed the high predictive ability of this
model. In addition, the results were visualized using a Streamlit-based interactive dashboard to enhance
the usability of the model for decision-making. These findings highlight the potential of XGBoost as a
powerful tool for fire risk classification and emphasize its relevance in supporting early warning
systems and evidence-based disaster mitigation policies in urban environments.

Application of K-Means Clustering for Regency/City Clustering in East Java Based on 2024 Human Development Index Indicators

Kholidatus Emilia — 2025-10-28

This study applies the K-Means clustering algorithm to group 38 regencies and cities in East Java Province based on five Human Development Index (HDI) indicators for the year 2024. These indicators include Life Expectancy (UHH), Expected Years of Schooling (HLS), Mean Years of Schooling (RLS), and Real Expenditure Per Capita (PPK). The aim of this research is to uncover hidden patterns and disparities in regional development, which can be used as a basis for more targeted and data-driven policy interventions.The optimal number of clusters was determined using three evaluation metrics: the Elbow Method, Silhouette Score, and Davies-Bouldin Index. These evaluations collectively identified three distinct clusters. Cluster 0 represents regions with high levels of development across all indicators. Cluster 1 consists of regions with moderate development levels and potential for improvement, while Cluster 2 contains regions with significantly lower values, particularly in education and income metrics.In addition to clustering, a correlation analysis was conducted to examine the relationship between HDI and its supporting indicators. The results show that Mean Years of Schooling (RLS) and Real Expenditure Per Capita (PPK) have the strongest positive correlation with HDI across all clusters. This highlights the key role of education and economic well-being in improving human development. The findings emphasize the importance of clustering analysis in shaping equitable and region-specific development strategies.