International Journal Of Computer Sciences and Mathematics Engineering

Analysis of Trends and Profitability Growth Opportunities of the VTuber Industry Using Exploratory Data Analysis (EDA) Methodology

Ananda Aulia — 2025-03-17

The VTuber industry has experienced significant growth in recent years, with VTubers as virtual influencers engaging with global audiences through live streaming and content creation. However, the financial performance of this industry remains underexplored. By applying EDA, this research investigates the key factors influencing the profitability of VTubers, including audience engagement, monetization strategies, and content diversity. The analysis is conducted using data collected from various VTuber channels, including viewer statistics, revenue streams (such as Super Chats, merchandise, and sponsorships), and social media metrics.

The EDA methodology proves to be effective in providing valuable insights into patterns and relationships between variables affecting profitability in this industry. Through data visualization and descriptive statistical analysis, EDA identifies factors contributing to VTuber performance, such as audience interaction levels and monetization success. The findings demonstrate that EDA offers a clear overview of trends and growth opportunities within the VTuber industry, helping content creators, investors, and industry stakeholders make more strategic, data-driven decisions.

Analysis of SVM Algorithm And SMOTE Technique on Public Sentiment With Google Maps Reviews Towards The Quality of Education at LP3I Across Indonesia

Maulian Saputra — 2025-03-28

This study analyzes public sentiment toward LP3I’s education quality in Indonesia using the Support Vector Machine (SVM) algorithm and the Synthetic Minority Over-sampling Technique (SMOTE). Reviews from Google Maps were collected, preprocessed, translated, and labeled using the Natural Language Toolkit (NLTK). The classification model was tested with and without SMOTE to address class imbalance. The results show that applying SMOTE improves the model’s ability to detect neutral and negative sentiments, increasing accuracy from 86.93% to 88.44% and significantly enhancing recall for minority classes. However, while SMOTE helps create a more balanced classification, it also introduces a potential risk of overfitting due to synthetic data generation. Overall, the integration of SVM and SMOTE improves sentiment classification performance, providing valuable insights for LP3I to enhance its academic and administrative services based on public perception.

E-COMMERCE APPLICATION FOR SALE OF FRAGILE GOODS AT THE WEBSITE-BASED BUSINESS STORE OF PATRAH

Yahya Muhaimin Sinaga — 2025-02-20

The development of information technology today is a necessity that must be mastered by every individual or organization to increase effectiveness and efficiency in a field, especially in business. There is a use of information technology to support a very diverse business, one of which is the use of electronic commerce or what is commonly called electronic commerce. Toko Usaha Patrah is a shop that sells household appliances. So far, the sales processing system has been less than optimal, and it is difficult to reach customers. Based on these problems, Usaha Patrah requires a good online sales business process to support the smooth running of the business. So the implementation of a web-based e-commerce application to expand its business and be known to many people.

ANALYSIS OF HEART FAILURE PREDICTION WITH RANDOM FOREST ALGORITHM AND LINEAR REGRESSION

Ismar Hidayat — 2025-03-03

Predicting the risk of heart failure is an important step in the prevention and early treatment of potentially fatal cardiovascular diseases. This study aims to compare the performance of two machine learning algorithms, namely Random Forest and Linear Regression, in predicting heart failure based on patient data that includes variables such as age, blood pressure, cholesterol levels, and other health history. The results show that the Random Forest algorithm is significantly superior in terms of prediction accuracy compared to Linear Regression, especially on data with a pattern of the number of data used. However, Linear Regression remains relevant in providing more stable results on differences in the amount of data used and has a more significant effect on the variables of heart failure. Therefore, a Random Forest-based prediction model is recommended to predict heart failure if it has a large amount of tranning data, and Linear Regression is recommended for prediction stability. The implementation of this model is expected to help medical practitioners in making more appropriate and accurate decisions to prevent the occurrence of heart failure in high-risk patients.

ANALYSIS OF THE SALES POTENTIAL OF BUMDES PRODUCTS USING THE K-MEANS CLUSTERING ALGORITHM

Arifin — 2025-03-03

This research aims to measure the strategy and potential of Village-Owned Enterprises (BUMDes) in Tandam Hulu II Village using Business Intelligence (BI) Tools supported by the K-Means Clustering method. BI Tools is used to analyze the performance and potential of BUMDes, while K-Means Clustering groups community data based on economic, social, and demographic characteristics. The results of the study show that BI Tools helps BUMDes accurately map the economic potential of villages and develop appropriate development strategies. Clustering analysis allows the identification of community groups with similar characteristics, so that BUMDes programs can be tailored to the specific needs of each cluster. This implementation also improves operational efficiency by focusing resources on high-potential sectors. In conclusion, the implementation of BI Tools and K-Means Clustering supports data-based decision-making in BUMDes, maximizes the economic potential of villages, and encourages sustainable economic growth.

SENTIMENT CLASSIFICATION ON E-COMMERCE USER REVIEWS WITH NATURAL LANGUANGE PROCESSING (NLP) AND SUPPORT VECTOR MACHINE (SVM) METHODS

Jimmy Iqbal Wiranata Siregar — 2025-03-24

This research aims to build a classification model that can categorize e-commerce user reviews into positive, negative, or neutral sentiments. By using NLP techniques to process the review text and SVM as a classification algorithm, it is expected that this model can provide high accuracy in determining user sentiment. Common words that do not contribute to sentiment analysis, such as “and,” “which,” “for”, are removed, and SVM is applied after the review data is transformed into vectors using the TF-IDF method. The SVM model will be trained using training data that has been labeled with sentiment.