Claim Missing Document
Check
Articles

Found 36 Documents
Search

Leveraging Various Feature Selection Methods for Churn Prediction Using Various Machine Learning Algorithms Kusnawi, Kusnawi; Ipmawati, Joang; Asadulloh, Bima Pramudya; Aminuddin, Afrig; Abdulloh, Ferian Fauzi; Rahardi, Majid
JOIV : International Journal on Informatics Visualization Vol 8, No 2 (2024)
Publisher : Society of Visual Informatics

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62527/joiv.8.2.2453

Abstract

This study aims to examine the effect of customer experience on customer retention at DQLab Telco, using machine learning techniques to predict customer churn. The study uses a dataset of 6590 customers of DQLab Telco, which contains various features related to their service usage and satisfaction. The data includes various features such as gender, tenure, phone service, internet service, monthly charges, and total charges. These features represent the demographic and service usage information of the customers. The study applies several feature selection methods, such as ANOVA, Recursive Feature Elimination, Feature Importance, and Pearson Correlation, to select the most relevant features for churn prediction. The study also compares three machine learning algorithms, namely Logistic Regression, Random Forest, and Gradient Boosting, to build and evaluate the prediction models. This study finds that Logistic Regression without feature selection achieves the highest accuracy of 79.47%, while Random Forest with Feature Importance and Gradient Boosting with Recursive Feature Elimination achieve accuracy of 77.60% and 79.86%, respectively. The study also identifies the features influencing customer churn most, such as monthly charges, tenure, partner, senior citizen, internet service, paperless billing, and TV streaming. The study provides valuable insights for DQLab Telco in developing customer churn reduction strategies based on predictive models and influential features. The study also suggests that feature selection and machine learning algorithms play a vital role in improving the accuracy of churn prediction and should be customized according to the data context.
Performance Analysis Of Machine Learning Algorithms Using The Ensemble Method On Predicting The Impact Of Inflation On Indonesia's Economic Growth Abdulloh, Ferian Fauzi; Aminuddin, Afrig; Rahardi, Majid; Harianto, Fetrus Jari
JOIV : International Journal on Informatics Visualization Vol 8, No 4 (2024)
Publisher : Society of Visual Informatics

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62527/joiv.8.4.2567

Abstract

The warning of a global recession expected in 2023 is currently the world's concern. Global financial institutions have raised interest rates to lower inflation, which has led to this problem. This study aims to evaluate the effect of interest rates and inflation on Indonesia's economic growth and compare the performance of machine learning models, specifically Random Forest and XGBoost, in analyzing the impact of inflation. A qualitative methodology was used for the literature survey, while the quantitative approach involved the implementation of machine learning algorithms using the Ensemble Method. The results show that Random Forest performs better than XGBoost in predicting the impact of inflation on economic growth, with MSE values of 0.799 and 0.864 and MAE of 0.576 and 0.619, respectively. In addition, the R-squared value of Random Forest 0.908 is also higher than that of XGBoost 0.901, indicating that the model can better explain the variation in the target data. The practical implication of this study is that the Random Forest model can be more effectively used in analyzing the impact of inflation on Indonesia's economic growth. Recommendations for future research include exploring other methods and using more extended time series to deepen the understanding of the relationship between interest rates, inflation, and economic growth.
Perbandingan Algoritma Support Vector Machine dan K-Nearest Neighbors Pada Sinyal Tubuh Perokok Musthofa, Alif; Rahardi, Majid
The Indonesian Journal of Computer Science Vol. 12 No. 6 (2023): The Indonesian Journal of Computer Science (IJCS)
Publisher : AI Society & STMIK Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33022/ijcs.v12i6.3290

Abstract

Merokok adalah kebiasaan yang sulit dihilangkan dalam masyarakat. Rokok mengandung bahan berbahaya dan bisa menyebabkan kanker serta penyakit pernapasan. Merokok juga meningkatkan risiko infeksi tuberkulosis. Perokok pasif yang terpapar asap rokok sangat berisiko bagi kesehatan. Penelitian ini bertujuan untuk melakukan evaluasi dan perbandingan antara algoritma Support Vector Machine (SVM) dan K-Nearest Neighbors (KNN) dalam klasifikasi sinyal tubuh perokok. Hasil evaluasi menunjukkan bahwa penggunaan SVM dengan kernel linear dan metode forward selection menghasilkan akurasi tertinggi sebesar 75%, yang melampaui akurasi tertinggi KNN sebesar 72%. Dari hasil tersebut penggunaan metode forward selection meningkatkan akurasi dibandingkan dengan penggunaan semuafitur yang tersedia, kecuali pada SVM dengan kernel RBF. Evaluasi pada penelitian ini menggunakan Confuntion Matrix dan Record klasifikasi. Adapun hasil kinerja model pada class “Tidak merokok” menggunakan SVM mendapatkan nilai presisi (84%), recall (75%), f-1 score(79%) dan KNN mendapatkan nilai presisi (75%), recall (83%), f-1 score(79%). Sedangkan pada class “Merokok” menggunakan SVM mendapatkan nilai presisi (64%), recall (75%), f-1 score(69%) dan KNN mendapatkan nilai presisi (64%), recall (53%), f-1 score(58%).
Detection of Ripeness in Oil Palm Fresh Fruit Bunches Using the YOLO12S Algorithm on Digital Images Nur'aini, Linnda Prawidya; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 4 (2025): August 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i4.10250

Abstract

Indonesia is the world's largest producer of palm oil, with a production volume reaching 46.82 million tons in 2022. This industry heavily relies on the quality of Fresh Fruit Bunches (FFB) harvests, which is determined by the accuracy of ripeness at the time of harvest. Unfortunately, ripeness assessment of FFB is still conducted manually and subjectively by field workers, posing risks to both efficiency and production accuracy. Although various studies have employed YOLOv5 and YOLOv8 for fruit ripeness detection, few have explored the potential of YOLO12s in classifying FFB ripeness in a comprehensive and efficient manner. In this study, we present the application of the YOLO12s algorithm to automatically classify the ripeness levels of oil palm FFB using digital images. The YOLO12s model was trained on 14,620 FFB images categorized into four ripeness levels: unripe, under-ripe, ripe, and overripe. Evaluation results showed a precision of 93.1%, recall of 95.9%, mAP@0.50 of 97.8%, and mAP@0.50–0.95 of 78.8%. The model was able to perform inference in approximately 4.7 milliseconds per image and demonstrated good generalization despite challenges related to varying lighting conditions. These results indicate that YOLO12s holds great potential to replace subjective manual methods with a more accurate, consistent, and efficient classification solution to support the harvesting process in the palm oil industry.
Comparative Study of Linear Regression, SVR, and XGBoost for Stock Price Prediction After a Stock Split Andrika, Muhammad Yusuf; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 4 (2025): August 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i4.10220

Abstract

This study aims to identify the most effective regression method for predicting the closing stock price of Bank Central Asia (BBCA) following the stock split event on October 12, 2021. Accurate post-split price predictions are crucial for helping investors comprehend new market dynamics, yet there is limited research evaluating the performance of regression models on BBCA’s stock after such corporate actions. Using data obtained through web scraping from the Indonesia Stock Exchange, this study tested three regression algorithms Linear Regression, Support Vector Regression, and XGBoost Regressor on post-split data. The selected input features were open_price, first_trade, high, low, and volume, while the target was close_price. The dataset was divided using an 80:20 train-test split and evaluated with RMSE, MAPE, and R-squared metrics. Results showed that Linear Regression achieved the best performance RMSE: 50.41, MAPE: 0.0048, R²: 0.9971, followed by XGBoost RMSE: 69.12, MAPE: 0.0058, R²: 0.9946, and SVR RMSE: 80.98, MAPE: 0.0069, R²: 0.9925. These findings indicate that BBCA’s post-split stock data exhibits a linear pattern, making Linear Regression the most suitable and efficient method. This suggests that simpler models can outperform more complex algorithms when applied to stable and structured financial datasets.
A Sentiment Analysis of Public Perception Toward Pets in Public Spaces Using Logistic Regression and Word Embedding Febianty, Dennita Noor; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 4 (2025): August 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i4.10245

Abstract

Addressing the complex social debate over pets in public areas, this study assesses public sentiment by analyzing a dataset of YouTube comments. We employed a machine learning pipeline beginning with data collection via the YouTube API, followed by rigorous text preprocessing and SMOTE-based class balancing for the training data. For classification, a Logistic Regression model was trained on contextual features generated by Word Embeddings (Word2Vec) and optimized through hyperparameter tuning. The final model proved highly effective, yielding a test accuracy of 92.74% with F1-scores of 0.84 for the negative class and 0.95 for the positive class. Ultimately, this research establishes an effective approach to measuring public opinion on social issues in Indonesia, providing actionable insights for public space administrators and policymakers.
Comparative Analysis of LightGBM and Random Forest for Daily Bitcoin Closing Price Prediction with Ensemble Approach Nolejanduma, Dionisius Nusaca Redegnosis; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 5 (2025): October 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i5.10589

Abstract

This study performs a comparative analysis of the LightGBM and Random Forest algorithms in predicting daily Bitcoin closing prices, with an exploration of an Ensemble approach for potential improvements in accuracy. A quantitative research design is employed, utilizing historical Bitcoin (BTC-USD) data from September 2015 to July 2025, enriched with various technical indicators. Data preprocessing, model training, and evaluation were carried out using Python in Google Colaboratory, with the dataset split into 80% for training and 20% for testing. Model performance was evaluated using the Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and the R-squared (R²) statistic, with statistical significance tests to ensure robust comparisons. A simple Linear Regression model was also included as a baseline. The findings reveal that Random Forest outperformed LightGBM, achieving an MAE of 11,599.74, an RMSE of 19,262.31, and an R² of 0.431, compared to LightGBM’s MAE of 12,285.42, RMSE of 19,995.04, and R² of 0.386. Although the Ensemble model showed slight improvements over LightGBM, it did not surpass Random Forest. The relatively low R² values across all models reflect the inherent volatility and difficulty in predicting Bitcoin prices. The study concludes that Random Forest demonstrates superior robustness for Bitcoin forecasting. Importantly, this work provides a novel empirical contribution by being one of the first to directly benchmark RF, LightGBM, and their Ensemble for Bitcoin prediction, highlighting that a simple averaging Ensemble does not guarantee superior performance. This finding provides a foundation for developing more refined Ensemble strategies tailored to high-volatility assets.
Comparative Analysis Transfer Learning Models for Early Detection of Pneumonia using Chest X-ray Images Rida, Rachmasari Annisa; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 5 (2025): October 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i5.10857

Abstract

Pneumonia is a serious respiratory disease that continues to be a major worldwide health issue, especially in nations that are struggling with limited medical resources. Early and accurate detection is essential to improve patient outcomes and reducing the rate of death. This study compares the performance of two deep learning architectures, DenseNet121 and ResNet50, using transfer learning for pneumonia detection from chest X-ray images. The dataset consists 5,856 images with two classes, NORMAL and PNEUMONIA, split into training 60%, validation 20%, and testing 20%. Pretrained ImageNet weights were used as fixed feature extractors, with a custom classification layers. Evaluation metrics included accuracy, precision, recall, F1-score, and confusion matrix. On the internal test set, DenseNet121 achieved 92% accuracy, with precision 0.79, recall 0.94, and F1-score 0.86 for NORMAL class, and precision 0.98, recall 0.91, and F1-score 0.94 for PNEUMONIA class. ResNet50 reached 81% accuracy, with precision 0.61, recall 0.80, and F1-score 0.70 for NORMAL class, and precision 0.92, recall 0.81, and F1-score 0.86 for PNEUMONIA class. External testing on an independent set of 200 images (100 images per class) yielded 98% accuracy for DenseNet121 and 85% for ResNet50. These results show that DenseNet121 provides better overall performance and lower false-negative risk for pneumonia cases, highlight the potential of DenseNet121 as a foundation for AI-assisted diagnostic tools in clinical practice.
Image-Based Classification of Indonesian Traditional Houses Using a Hybrid CNN-SVM Algorithm Ikhsan, M.; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 5 (2025): October 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i5.10864

Abstract

The diversity of Indonesian traditional houses represents a cultural heritage that must be preserved. However, the lack of interest among younger generations and the difficulty in recognizing the distinctive architectural characteristics of traditional houses present challenges to preservation efforts. This study aims to develop an image classification model for Indonesian traditional houses using a hybrid CNN-SVM approach to improve recognition accuracy. The dataset consists of 3,919 images from five classes of traditional houses, namely gadang, joglo, panjang, tongkonan, and honai, with an 80% training split, 10% validation, and 10% testing. The data were processed through resizing, augmentation, and normalization before being trained using a CNN architecture with five convolutional layers as a feature extractor and an SVM serving as a multi-class classifier. The experimental results show that the hybrid CNN-SVM model achieved an accuracy of 96.68%, with consistently high precision, recall, and F1-score across all classes. These findings demonstrate that integrating CNN as a feature extractor and SVM as the final classifier can enhance the model’s generalization capability in distinguishing images of Indonesian traditional houses.
Sentiment Analysis of Economic Policy Comments on YouTube Using Ensemble Machine Learning Nandini, Kety; Rahardi, Majid
Journal of Applied Informatics and Computing Vol. 9 No. 5 (2025): October 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i5.10897

Abstract

Public sentiment analysis of economic policies is increasingly important in the digital age, as social media platforms have become the main arena for public discussion. This study analyzes YouTube comments related to Tom Lembong's economic policies to address the lack of policy sentiment analysis tools in Indonesian. A dataset containing 1,029 comments was collected and systematically processed using normalization, stop word removal, and stemming techniques tailored to Indonesian. To overcome data scarcity and class imbalance, advanced data augmentation methods—synonym replacement, random insertion, and random deletion—were applied, expanding the dataset to 2,169 samples. Feature extraction used TF-IDF vectorization (unigram, bigram, trigram) and CountVectorizer, followed by an 80:20 split into training and testing sets. Several machine learning algorithms, including Support Vector Machine (SVM), Logistic Regression, Random Forest, Gradient Boosting, and Naïve Bayes, were evaluated with hyperparameter tuning through grid search. The results showed that SVM with TF-IDF bigrams achieved the best performance (accuracy: 96.08%, F1-score: 96.03%). Class-level evaluation showed high performance for negative sentiment (F1-score: 0.97) and positive sentiment (F1-score: 0.97), while neutral sentiment was more challenging (F1-score: 0.90) due to ambiguity, sarcasm, and fewer samples. The ensemble model, which combines several optimized SVM variants with soft voting, achieved robust and stable performance (accuracy and F1-score: 95.16%). These findings confirm the effectiveness of the ensemble-based approach for Indonesian sentiment analysis, while providing valuable insights into public perceptions of economic policy in the digital space.