Claim Missing Document
Check
Articles

Found 5 Documents
Search

Klasifikasi Gaya Belajar VARK Siswa Sekolah Dasar Menggunakan K-Nearest Neighbor dan Naive Bayes Berbasis Kuesioner: Classification of VARK Learning Styles of Elementary School Students Using K-Nearest Neighbor and Questionnaire-Based Naive Bayes Deden Adi Mardian Lesmana; Fathoni Mahardika; Dani Indra Junaedi
Jurnal Informatika dan Multimedia Vol. 18 No. 1 (2026): Jurnal Informatika dan Multimedia
Publisher : Politeknik Negeri Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33795/jtim.v18i1.9441

Abstract

Gaya belajar memiliki peranan penting dalam menentukan bagaimana siswa memahami informasi, khususnya pada jenjang sekolah dasar yang berada pada tahap perkembangan berpikir konkret. Penelitian ini bertujuan memetakan preferensi belajar siswa menggunakan model VARK (Visual, Aural, Read/Write, Kinesthetic) serta membangun model klasifikasi berbasis algoritma K-Nearest Neighbor (KNN). Sebanyak 40 siswa kelas IV–VI berpartisipasi dengan mengisi 16 item kuesioner VARK. Data diolah melalui proses pembersihan, normalisasi Min–Max, dan pembentukan fitur empat dimensi. Model KNN diuji menggunakan variasi nilai k melalui Stratified 5-Fold Cross Validation. Hasil penelitian menunjukkan bahwa kategori Kinesthetic dan Visual merupakan preferensi belajar yang paling dominan di SDN Cipatat. Model KNN memberikan performa terbaik pada k = 5, dengan akurasi rata-rata 82%, precision 0,81, recall 0,80, dan F1-score 0,79. Analisis confusion matrix memperlihatkan bahwa kategori Kinesthetic dan Visual lebih mudah diprediksi, sementara Aural dan Read/Write memiliki tumpang tindih fitur yang lebih besar. Temuan ini menunjukkan bahwa pendekatan berbasis data dapat memberikan gambaran objektif mengenai preferensi belajar siswa serta mendukung strategi pembelajaran yang lebih adaptif.
Analisis Kebutuhan pada Perancangan Antarmuka Pengguna Aplikasi Koperasi Menggunakan Design Thinking Fina Nur'aeni; Fathoni Mahardika; Dani Indra Junaedi; Agun Guntara
TAMIKA: Jurnal Tugas Akhir Manajemen Informatika & Komputerisasi Akuntansi Vol 5 No 2 (2025): TAMIKA: Jurnal Tugas Akhir Manajemen Informatika & Komputerisasi Akuntansi
Publisher : Universitas Methodist Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46880/tamika.Vol5No2.pp250-254

Abstract

The development of information technology encourages cooperatives to adapt to digital systems to improve efficiency, transparency, and accuracy in managing membership and financial data. This study aims to analyze user needs as the basis for designing a user interface for a cooperative membership and financial management application using the Design Thinking method. This method consists of five stages: empathize, define, ideate, prototype, and test. The study focused on needs analysis with the initial stages: empathize, define, and ideate. The empathize stage involved interviews with cooperative administrators and members, followed by identification through the define stage. The analysis results indicated that users require an efficient, secure, and transparent digital system with key features such as member registration, online payments, bill notifications, and automated financial reports. The ideate stage resulted in a design idea that comprehensively describes user needs. This allows for a design that adapts to the analysis results. Thus, the Design Thinking approach has proven effective in understanding user problems and needs and provides a strong foundation for designing a cooperative application interface that is user-friendly and appropriate to the organization's operational context.
Analisis Kebutuhan Pengguna untuk Perancangan Antarmuka Aplikasi Layanan Pengantaran Menggunakan User-Centered Design Nida Shofwatunnisa; Fathoni Mahardika; Dani Indra Junaedi; Agun Guntara
TAMIKA: Jurnal Tugas Akhir Manajemen Informatika & Komputerisasi Akuntansi Vol 5 No 2 (2025): TAMIKA: Jurnal Tugas Akhir Manajemen Informatika & Komputerisasi Akuntansi
Publisher : Universitas Methodist Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46880/tamika.Vol5No2.pp350-355

Abstract

Delivery services are one type of service that is widely used by the public to facilitate the process of delivering goods and orders. However, the current delivery system is still not running optimally because the use of digital technology has not been maximized. This study aims to analyze user needs in the design of a delivery service application using the User-Centered Design (UCD) method. The UCD approach was chosen because it focuses on users by placing their needs and experiences as the main aspects in the system design process. This study was conducted up to the second stage, which was to understand the context of use and identify user needs. Data collection was carried out by distributing questionnaires to ten respondents consisting of customers, business owners, and couriers. The results of the study show that users need key features such as service ordering, real-time delivery tracking, automatic notifications, and direct communication between customers and couriers. In addition, users also want an application that is easy to use, secure, and has stable performance. The results of this analysis form the basis for the application design in the next stage to produce a system that meets user needs.
Implementasi Algoritma Random Forest untuk Prediksi Volume Pengunjung pada Sektor Layanan Publik (Studi Kasus Unit Layanan Publik di Jawa Barat Tahun 2021-2022) Anisa Pebriyani Huslan; Fathoni Mahardika; Dani Indra Junaedi
Riau Jurnal Teknik Informatika Vol. 4 No. 3 (2025): November 2025
Publisher : Prodi Teknik Informatika Universitas Pasir Pengaraian

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30606/rjti.v4i3.4042

Abstract

Peramalan volume pengunjung harian yang akurat merupakan kunci untuk penjadwalan petugas yang andal, antrean yang singkat, dan pengalaman warga yang lebih baik pada layanan publik. Penelitian ini mengajukan pendekatan praktis dan mudah direplikasi yang hanya memanfaatkan sinyal kalender untuk meramalkan kedatangan harian menggunakan Random Forest. Dataset harian unit pelayanan publik tahun 2021 hingga 2022 berisi 730 observasi dan dibagi secara kronologis menjadi 80 persen pelatihan dan 20 persen pengujian agar menyerupai penerapan nyata serta menghindari kebocoran informasi. Prediktor mencakup hari, bulan, hari dalam minggu, indikator akhir pekan, minggu dalam bulan, serta penanda awal dan akhir bulan. Model menggunakan 200 pohon dengan max_depth = 8. Kinerja dievaluasi menggunakan MAE, MAPE, RMSE, dan R² serta dibandingkan dengan dua baseline yang kuat, yaitu naive lag-1 dan rata-rata hari dalam minggu. Pada data uji, model mencapai R² = 0.873, RMSE = 16.328, MAE = 13.103, dan MAPE sekitar 5,00 persen, melampaui kedua baseline. Secara kuantitatif, model ini menurunkan kesalahan prediksi (MAPE) masing-masing sekitar 64% dan 67% dibandingkan baseline naïve lag-1 dan rata-rata hari dalam minggu, menegaskan keunggulan empirisnya. Analisis feature importance menempatkan hari dalam minggu dan akhir pekan sebagai variabel paling berpengaruh dan sejalan dengan intuisi operasional. Temuan ini menunjukkan bahwa peramalan berbiaya rendah dan hemat data mampu memberikan akurasi yang siap pakai untuk penjadwalan, pemantauan risiko antrean, dan alokasi kapasitas, serta mudah ditingkatkan dengan konteks tambahan seperti hari libur dan cuaca.
Algoritma Linear Search dan Binary Search Berdasarkan Ukuran dan Kondisi Keterurutan Data Gilang Abdul Rahman; Dani Indra Junaedi; Deris Santika
Infoman's : Jurnal Ilmu-ilmu Informatika dan Manajemen Vol. 19 No. 2 (2025): Infoman's
Publisher : LPPM & Fakultas Teknologi Informasi UNSAP

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This study compares the performance of Linear Search and Binary Search in data retrieval under varying dataset sizes and ordering conditions. Rather than relying solely on theoretical complexity, Binary Search is evaluated end-to-end by including the required sorting step prior to searching. A quantitative experiment is conducted in Python 3.13.9 using shuffled unique integer arrays with sizes ranging from to . Four target scenarios are tested: target located at the beginning, middle, end, and target not found. The primary metrics are execution time and summary statistics (median and mean) computed from repeated runs for each scenario. The results indicate that for a single search on initially unsorted data, the Sorting+Binary approach tends to yield a higher total time than Linear Search because sorting dominates the overall cost, while the binary search component itself remains comparatively small. The contribution of this work is an end-to-end evaluation that accounts for sorting overhead and provides practical guidelines for selecting the appropriate search algorithm across dataset sizes and query scenarios. These findings highlight that algorithm selection should account for data characteristics and preprocessing overhead; Binary Search is most beneficial when data is already sorted or when sorting costs can be amortized across repeated queries.