Journal of Student Research Exploration

The Comparison of Statement Analysis on Disney Case Using Naive Bayes, SNM, and Logistic Algorithm Methods

Rizkiyanti Choirunnisa, Budi Prasetiyo — Mon, 13 Apr 2026 00:00:00 +0000

The Walt Disney Company or also known as Disney, is one of the most famous companies in the world that focuses on the production of animation and film. Disney has been serve in this entertainment industry for over 90 years. Since Disney’s first film was released, Disney was become very famous until this day, especially when Disney has collaborates with many companies, it’s not only focusing on animation production but also making films and many live-action versions of the animations. Recently, Disney has been a hot topic among Disney’s movie fans due to the selection of actors for live action movie characters. Therefore, on this time, the author will conduct a sentiment analysis of Disney using analysis methods called Naïve Bayes, Support Vector Machine, and Logistic Algorithm. After passing through all the testing stages, the highest accuracy result is analysis using the Support Vector Machine (SVM) method. The test results show a high accuracy of 0.60 or 60% with the highest F1-Score in the Positive sentiment class by 67%.

Implementation of Lexicon-Based and SVM Methods in Sentiment Analysis of Sayurbox App Users

Raihan Muhammad Rizki Rahman, Budi Prasetiyo — Mon, 13 Apr 2026 00:00:00 +0000

The ever-growing technology certainly produces a large amount of data, which can provide useful information if analyzed and used properly. The purpose of this research is to analyze user sentiment towards the Sayurbox application on the Google Play Store with a Lexicon-Based approach and the Support Vector Machine (SVM) algorithm. User review data is obtained through web scraping with a total of 16,468 reviews. After preprocessing and sentiment labeling, training and test data were divided. The results showed that SVM achieved accuracy, recall, and precision of 94%, 96%, and 96% respectively, with 9 prediction errors. The model tends to predict reviews as positive sentiment, indicating user satisfaction with Sayurbox's product service, delivery, quality, and price. The findings make a contribution to the understanding of user sentiment in e-commerce services and can assist Sayurbox in improving their user experience.

Clustering Analysis of Customers Based on Purchasing Patterns with K-Means Clustering

Wayne Joel Marcelino Lubis — Mon, 13 Apr 2026 00:00:00 +0000

There are various techniques to classify data, one of which is clustering. What distinguishes clustering techniques from classification techniques is that they do not rely on the labels in the dataset. The main purpose of clustering is to divide data into several clusters based on similar characteristics, while Classification Technique is a technique of grouping data based on the similarity of the labels of the data under study. In this study, the dataset was created using secondary data from kaggle. The analysis process begins with data pre-processing to normalize the variables used, followed by the application of the K-Means Clustering method to group customers into several clusters based on the similarity of their purchasing patterns. This research demonstrates the potential of using clustering analysis to improve understanding of customer behavior and develop more effective business strategies.

Integrating Convolutional Neural Network Features Extraction with Extreme Learning Machines for Image Classification of Pandava Characters in Wayang Kulit

Alfiatul Fitria, Budi Prasetiyo — Tue, 14 Apr 2026 00:00:00 +0000

This research focuses on the utilization of image processing techniques—the Convolutional Neural Networks (CNNs) and Extreme Learning Machine (ELMs)—to classify the characters of Wayang Kulit automatically. The Pandava characters or casts are classified in accordance with the characters from traditional Indonesian puppets, commonly known as shadow puppets. The focus is to introduce such rich cultural heritage to younger generations by using technology. Prior research has utilized classification of characters using Convolutional Neural Networks(CNNs), Extreme Learnings Machines(ELMs), and Support Vector Machines(SVMs), which led to varied accuracy levels. In our subsequent experiments, three proposed models, with varying underlying model assumptions, were evaluated. The proposed models generated moderate accuracies ranging from 39 to 52%. The results suggest that our models have room for further development to enhance their performance. Strategies from parameter tuning to the in-depth analysis of the confusion matrix are discussed. Above all, the research is geared towards ensuring the appreciation and preservation of traditional cultural heritage in this digital era.

Enhanced Out-of-Fold Stacking with Feature Grouping and Model-Specific Transformations for Diabetes Prediction Improvement

Thu, 16 Apr 2026 00:00:00 +0000

Diabetes mellitus is a chronic disease with serious implications for global health. Early detection is essential to reduce these risks, and machine learning methods are widely used in diabetes prediction. However, improving accuracy remains a major challenge in the development of predictive models. This study proposes a stacking-based ensemble learning approach with an out-of-fold (OOF) scheme to improve classification performance. The proposed method consists of several systematic steps, namely (1) data preprocessing via median imputation of invalid values and feature transformation according to model characteristics, (2) the creation of base learners comprising Logistic Regression, Gaussian Naïve Bayes, Support Vector Machine, Random Forest, and XGBoost, (3) model training using Stratified Cross Validation 5 Fold to generate OOF predictions, (4) combining all OOF predictions into a meta-feature matrix, and (5) training an XGBoost-based meta-model to generate the final prediction. This approach enables the meta-model to optimally learn the relationships among the outputs of the baseline models. Experimental results show that the proposed method achieves an accuracy of 91.15%, precision of 90.65%, recall of 83.21%, and an F1-score of 86.77%. These results indicate that stacking is effective in improving the accuracy of diabetes predictions.