JDMIS: Journal of Data Mining and Information Systems

Analisis Sentimen Ulasan Aplikasi Maxim Merchant dengan Support Vector Machine (SVM) dan Random Forest

2025-06-17T02:43:53+00:00

The development of digital technology, especially mobile devices, has led to an increase in application-based services. One important aspect in app development is to deeply understand user perception and satisfaction. This study aims to analyze user sentiment towards the Maxim Merchant application based on reviews obtained from the Google Play Store platform. A total of more than 2800 Indonesian-language reviews were collected using web scraping techniques. The review data was processed through pre-processing stages such as text cleaning, normalization, tokenization, removal of unimportant words, and stemming. Sentiments are categorized into positive and negative based on the review score, where scores of 1 to 3 are considered negative, and scores of 4 and 5 are considered positive. Word cloud visualization is used to show the dominant words of each sentiment category. The data is then converted into numerical form using TF-IDF and selected using the Chi-Square method. Classification was performed using Support Vector Machine and Random Forest algorithms. The evaluation results show that the Support Vector Machine algorithm performs better in classifying sentiment, especially in handling high-dimensional text data.

Klasifikasi Sentimen Ulasan Pengguna Aplikasi Qpon dengan Support Vector Machine dan Logistic Regression

2025-06-08T13:47:51+00:00

The increasing number of user reviews in mobile applications is an important source of information in understanding user satisfaction and experience with the services used. One of the applications used in this study is the Qpon application. Reviews left by users often contain positive or negative opinions that can influence other users in making decisions. Therefore, sentiment analysis is needed to determine the tendency of opinions in these reviews. This study aims to classify Qpon application user reviews into two sentiment categories, namely positive and negative. Data were collected through the web scraping method and obtained 866 review data. After going through text preprocessing stages such as removing unimportant words, normalization, and tokenization, the data were analyzed using the TF-IDF method as a feature representation, then classified using the Logistic Regression and Support Vector Machine (SVM) algorithms. The testing process was carried out using the Stratified K-Fold Cross Validation technique and measured based on five evaluation metrics, namely accuracy, precision, recall, F1-score, and ROC AUC. The results showed that SVM had the highest accuracy and precision values, while Logistic Regression was superior in recall and ROC AUC. These findings indicate that SVM is superior in terms of classification accuracy, while Logistic Regression is more sensitive to positive reviews. This study is expected to be used as a reference for the development of a sentiment analysis system to improve application services based on user review data.

Developing Business Intelligence Dashboard for Sales KPI Monitoring in Advertising Agency: A Human-Centered Design Approach

2026-01-06T02:22:27+00:00

Digital advertising agencies in South Jakarta face significant challenges in monitoring sales performance due to data fragmentation across multiple platforms such as CRM, spreadsheets, and digital advertising tools. Conventional manual reporting processes lead to data latency, high error rates, and delayed strategic decision-making. This study aims to develop a Business Intelligence (BI) dashboard to monitor Sales Key Performance Indicators (KPIs) in real-time, utilizing a Human-Centered Design (HCD) approach to ensure high usability and adoption. The research methodology follows the ISO 9241-210 standard for HCD, encompassing four iterative phases: understanding the context of use, specifying user requirements, producing design solutions, and evaluating designs. The system was developed using Google Looker Studio with a data warehouse architecture integrating Google BigQuery. Testing was conducted involving 15 internal stakeholders using the System Usability Scale (SUS) and User Experience Questionnaire (UEQ). The results demonstrated a SUS score of 82.5 (Excellent) and positive benchmarks in efficiency and perspicuity metrics. The implementation of the dashboard reduced reporting time by 60% and improved data accessibility for executive decision-making. This study contributes to the literature by demonstrating how HCD principles can bridge the gap between technical BI capabilities and end-user cognitive needs in the creative industry context.

Implementasi Algoritma K-Means Dengan Normalisasi Min-Max Pada Analisis Data Ketidakbersekolahan Anak

2026-01-30T11:29:48+00:00

Anak-anak yang tidak bersekolah merupakan suatu masalah dalam dunia pendidikan yang masih menjadi tantangan, terutama di kalangan masyarakat dengan ekonomi rendah. Tingginya jumlah anak yang tidak mengenyam pendidikan dapat mengurangi kualitas sumber daya manusia dan memperbesar kesenjangan sosial. Penelitian ini bertujuan untuk mengkaji ketidakbersekolahan pada anak berdasarkan level pendidikan dan kelompok pengeluaran, dengan menggunakan pendekatan data mining. Metode yang diterapkan mencakup normalisasi Min-Max sebagai langkah awal dalam memproses data serta algoritma K-means Clustering untuk proses pengelompokan. Normalisasi Min-Max digunakan untuk menyamakan skala data dalam rentang 0 hingga 1, sehingga setiap variabel memiliki peran yang seimbang dalam perhitungan jarak. Data yang digunakan adalah data angka anak tidak sekolah Tahun 2023, yang mencakup tingkat pendidikan SD, SMP, dan SMA rentang kelompok pengeluaran dari kuantil 1 hingga 5. Temuan penelitian ini menunjukkan bahwa algoritma K-Means dengan k = 3 dapat mengelompokkan data menjadi tiga kluster utama, yakni tingkat ketidakbersekolahan yang tinggi, sedang, rendah. Ini mengindikasikan adanya hubungan antara level pengeluaran dan partisipasi anak dalam pendidikan.

Penerapan Normalisasi Data pada Angkatan Kerja Indonesia Bulan Februari 2025 Berdasarkan Kelompok Umur

2026-01-21T14:28:22+00:00

Data normalization is a crucial initial step in the data mining process, aiming to reduce scale differences in numerical attributes, allowing for more objective and accurate analysis. This study aims to implement and evaluate data normalization techniques on the Indonesian workforce in February 2025 based on age category. The data used is secondary data obtained from the Central Bureau of Statistics (BPS) thru the National Labor Force Survey (SAKERNAS), which includes numerical attributes such as the number of employed people, the number of unemployed, the size of the labor force, and the percentage of the working population. The normalization methods used in this study consist of Min-Max Normalization, Z-Score Normalization, and Decimal Scaling Normalization. The research process includes data collection, selection of data from the period February 2025, data cleaning, application of normalization techniques, and analysis of the normalization results. The research findings indicate that all three normalization methods successfully leveled the value scales across attributes that previously showed significant differences in their value ranges. Min-Max normalization is effective in converting data to a specific range, Z-Score can identify deviations from the mean value, while Decimal Scaling facilitates proportional comparisons between age categories. Empirically, this study confirms that the 25-44 age group will be the most dominant in the structure of the Indonesian workforce in February 2025. Implementing data normalization has proven to improve data quality and support more accurate labor analysis.

Prediksi Diabetes Berbasis Decision Tree Dengan Menggunakan Dataset Pima Indians Diabetes

2026-02-02T08:41:23+00:00

Diabetes mellitus is a chronic disease characterized by increased blood glucose levels and can lead to various serious complications if not treated early. This research aims to predict diabetes using the Decision Tree algorithm with the Pima Indians Diabetes dataset. The research stages include data processing, forming a Decision Tree model using the entropy criterion, and evaluating model performance. The results show that the model achieved an accuracy of 76.62%. Testing through a confusion matrix produced 83 True Negative samples, 35 True Positive samples, 16 False Positive samples, and 20 False Negative samples. The Glucose attribute was found to be the most dominant factor in the diagnosis, followed by BMI and Age. The resulting model is able to form clear and easy-to-understand decision rules so that it can be used as a decision support system in the early diagnosis of diabetes.

Klasifikasi Gempa Bumi Menggunakan Algoritma Decision Tree Berbasis Data BMKG

2026-02-02T08:41:51+00:00

This study was conducted to classify earthquakes using the Decision Tree algorithm based on data from the Meteorology, Climatology, and Geophysics Agency. Indonesia is a region with high seismic activity, which requires a systematic classification method to group earthquakes according to their characteristics. The data used in this study consisted of earthquake magnitude and depth parameters, which were classified into light, moderate, and strong earthquake classes. The research stages included data collection, data preprocessing, determination of earthquake classes, construction of a classification model using the Decision Tree algorithm, and evaluation of the classification results. The results showed that the Decision Tree algorithm was able to classify earthquakes effectively based on the combination of magnitude and depth values. The resulting model generated clear and easily interpretable decision rules to distinguish between light, moderate, and strong earthquake classes. The conclusion of this study indicated that the Decision Tree algorithm could be used as an effective and interpretable method for earthquake classification based on data from the Meteorology, Climatology, and Geophysics Agency.

Analisis Kualitas Layanan Terhadap Loyalty Behavior ada Aplikasi SRIBU Menggunakan Metode E-Servqual

2026-01-31T08:35:05+00:00

The phenomenon of low app ratings and technical complaints on the Google Play Store became the background of this main research to expand which service quality dimensions are able to maintain the user base. The method used is E-SERVQUAL which includes seven dimensions: Efficiency, System Availability, Fulfillment, Privacy, Responsiveness, Contact, and Compensation. Data were collected through questionnaires from 405 respondents who use the SRIBU Mobile application and analyzed using the Partial Least Square-Structural Equation Modeling approach through SmartPLS software. The results showed that the aggregate service quality was assessed as good with an average value of 1.0379. The results of hypothesis testing confirmed that the dimensions of Efficiency, System Availability, Fulfillment, Privacy, Responsiveness, and Contact have a significant effect on user satisfaction. Furthermore, user satisfaction was proven to have a very strong positive and significant influence on loyalty behavior with a path coefficient value of 0.875. However, the Compensation dimension was found to have no significant influence on satisfaction. In addition, the Contact and Compensation dimensions showed poor values indicating that there are aspects of the service that have not met user expectations. This study recommends that PT SRIBU Digital Kreatif prioritize improvements to customer support channels and compensation policies to minimize service failures and strengthen user loyalty amidst intense digital economic competition.