Rizal Rizal
Teknik Informatika, STMIK Widuri Jakarta

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

OPTIMALISASI KLASIFIKASI UJI EMISI SEPEDA MOTOR MENGGUNAKAN ALGORITMA NAÏVE BAYES Irwansyah Irwansyah; Rizki Dittyata; Rizal Rizal; Wiyono Wiyono; Firman Noor Hasan
Infotech: Journal of Technology Information Vol 10, No 2 (2024): NOVEMBER
Publisher : ISTEK WIDURI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37365/jti.v10i2.327

Abstract

Dense urban areas with high levels of industrial and transportation activity result in increased air pollutant emissions that threaten air quality and the health of their residents. The issue is the lack of utilization and optimization of motorcycle emission test classification through a machine learning approach. This research aims to utilize motorcycle emission test data and to determine the accuracy, precision, and recall results of the naive Bayes algorithm. The number of datasets used by the researchers is 2409 data points. Based on this data, it is divided into two parts: training data consisting of 1927 data points (80%) and testing data consisting of 482 data points (20%). The results of the motorcycle emission test data can be utilized for classification optimization, and the naive Bayes algorithm can be applied to classify and analyze the accuracy, precision, and recall results of the motorcycle emission test data. The accuracy result is 91.49%, the precision result for the pass classification is 93.72%, and the precision result for the fail classification is 83%, while the recall result for the pass classification is 95.47% and the recall result for the fail classification is 77.57%.      ABSTRAK Daerah perkotaan yang padat penduduk dengan tingkat aktivitas industri dan transportasi yang tinggi mengakibatkan peningkatan emisi polutan udara yang mengancam kualitas udara dan kesehatan warganya. Permasalahan belum  adanya pemanfaatan dan optimalisasi klasifikasi uji emisi sepeda motor melalui pendekatan machine learning. Penelitian ini bertujuan untuk memanfaatkan data uji emisi sepeda motor dan untuk mengetahui  hasil akurasi, presisi, dan recall dari algoritma naïve bayes. Adapun jumlah dataset yang peneliti gunakan sebanyak 2409 data. Berdasarkan data tersebut dibagi menjadi dua yaitu data training sebanyak 1927 data (80%) dan data testing sebanyak 482 data (20%). Hasil penelitian data uji emisi sepeda motor dapat dimanfaatkan untuk optimalisasi klasifikasi dan algoritma naïve bayes dapat diterapkan dalam mengklasifikasi dan menganalisis hasil akurasi, presisi, dan recall dari data uji emisi sepeda motor. Adapun hasil akurasinya sebesar sebesar 91,49%, hasil precision klasifikasi lulus sebesar 93,72% dan hasil precision klasifikasi tidak lulus sebesar 83%, dan hasil recall klasifikasi lulus sebesar 95,47% dan hasil recall klasifikasi tidak lulus sebesar 77,57%.
PENGEMBANGAN SISTEM INFORMASI PRESENSI BERBASIS ANDROID DENGAN TEKNOLOGI PENGENALAN WAJAH DAN GEOLOKASI UNTUK OPTIMALISASI PENGELOLAAN KEHADIRAN KARYAWAN Rendi Widjaya; Asrul Sani; Rizal Rizal
EBID: Ekonomi Bisnis Digital Vol 2, No 2 (2024): Desember
Publisher : STMIK Widuri Jakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37365/ebid.v2i2.332

Abstract

This study aims to develop an Android-based attendance information system by integrating facial recognition and geolocation technologies to enhance the accuracy and efficiency of employee attendance management. Facial recognition technology offers advantages in preventing fraud, such as proxy attendance, while geolocation ensures attendance can only be recorded at predefined locations. The research employs the Research and Development (R&D) method, including stages such as needs analysis, system design, application development, testing, and implementation. The results show that facial recognition achieves an average accuracy of 85.75%, with reduced performance under low-light conditions or when employees wear masks. The geolocation feature achieves an accuracy of 91.5% within a 30-meter radius, minimizing attendance from unauthorized locations. Attendance time per employee decreased from 50 seconds (manual) to 25 seconds (digital), improving efficiency by 50%. Employee attendance increased from 85% to 95%, with 100% elimination of proxy attendance cases. This system significantly improves attendance management, providing a more effective, transparent, and accurate process that positively impacts employee productivity and discipline.