Jurnal Data Mining dan Sistem Informasi

Business Intelligence for Food Security Decision-Making from Transformed Paddy and Rice Production Data in North Kalimantan

Arwan Arwan — 2026-08-15

The agribusiness sector in North Kalimantan Province faces challenges of supply chain inefficiencies and limited post-harvest infrastructure, which hinder food security and stability. This study aims to transform historical data on rice and paddy production from the Statistics Indonesia into Business Intelligence tools to support precise decision-making. The method used is a descriptive-quantitative approach, with an action research design. Historical data for the period 2018-2025 is processed and visualized in Looker studio. The main focus of the analysis is the extraction of production variables into technical indicators of efficiency, using yield percentages and geospatial mapping. The study's results show that the system successfully integrates production metrics, with an average actual yield of 59.15%, which is below the national standard of 64.02%. Analysis of the time-series chart shows a decrease in production, especially in Nunukan, which is a producer of Adan rice with premium varieties. Geospatial visualization reveals the region's polarization, where the administrative area is not directly proportional to production volume. The conclusion of this study confirms that the development of this dashboard is effective as a visual audit tool for detecting technical inefficiencies and spatial disparities. The transformation of tabular data into interactive insights enables policymakers to determine more targeted agricultural technology interventions to strengthen regional agribusiness supply chains.

PENERAPΑΝ ΜΕΤODE K-MEANS CLUSTERING UNTUK PENGELOMPOKAN FILM BERDASARKAN RATING DAN DURASI

pandega arga — 2026-08-15

Banyaknya jumlah film yang tersedia di berbagai platform membuat proses pengelompokan film berdasarkan karakteristiknya menjadi penting, baik untuk kebutuhan rekomendasi maupun analisis tren perfilman. Penelitian ini bertujuan untuk menerapkan algoritma K-Means Clustering dalam mengelompokkan film berdasarkan dua atribut, yaitu rating IMDb dan durasi film. K-Means merupakan salah satu algoritma clustering non-hierarki yang bekerja dengan mengelompokkan objek ke dalam k kelompok berdasarkan kedekatan jarak Euclidean terhadap titik pusat (centroid) tertentu. Penelitian ini menggunakan enam sampel film dengan rating tinggi (8,5 ke atas) yang dikelompokkan ke dalam tiga cluster (k=3). Data dinormalisasi menggunakan metode min-max sebelum diproses agar kedua atribut memiliki skala yang setara. Hasil penelitian menunjukkan bahwa algoritma K-Means berhasil mengelompokkan film ke dalam tiga cluster yang berbeda karakteristik durasinya, yaitu cluster film berdurasi pendek-sedang, cluster film dengan rating sangat tinggi berdurasi sedang, dan cluster film epik berdurasi sangat panjang. Proses iterasi konvergen setelah dua kali iterasi, menunjukkan bahwa algoritma K-Means efektif dan efisien digunakan untuk pengelompokan data film berskala kecil.

KLASIFIKASI TINGKAT STRES MAHASISWA BERDASARKAN AKTIVITAS HARIAN MENGGUNAKAN RANDOM FOREST

Vivi Aldani — 2026-08-15

Student mental health is increasingly recognized as a critical factor affecting academic performance and overall well-being. This study implements the Random Forest algorithm to classify student stress levels into three categories—Low, Medium, and High—based on daily activity survey data collected from 30 students at Politeknik Negeri Medan. The dataset comprises 13 input features covering academic pressure, lifestyle habits, gadget usage, and emotional condition, along with a self-reported stress label. Data preprocessing included label encoding of categorical responses and an 80:20 train-test split. The Random Forest model was built using 100 decision trees with the Gini Impurity criterion and sqrt max features. Evaluation on the test set of 6 samples yielded 100% accuracy with perfect precision, recall, and F1-score across all three classes. Feature importance analysis revealed that deadline pressure (0.142), academic workload (0.138), and difficulty concentrating (0.121) are the most influential predictors of student stress. The majority of respondents (66.67%) fell into the Medium stress category, underscoring the pervasiveness of academic pressure. These findings demonstrate the effectiveness of Random Forest for multi-class stress classification from survey data and highlight actionable targets for campus well-being interventions.

Analisis Pengaruh Durasi Penggunaan Media Sosial Terhadap Prestasi Akademik Mahasiswa Menggunakan Metode Simple Linear Regression

syalwa juwita — 2026-08-15

Media sosial telah menjadi bagian penting dalam kehidupan mahasiswa dan digunakan untuk
berbagai tujuan, baik akademik maupun nonakademik. Intensitas penggunaan media sosial yang
tinggi sering dikaitkan dengan perubahan perilaku belajar yang berpotensi memengaruhi prestasi
akademik. Penelitian ini bertujuan untuk menganalisis hubungan antara durasi penggunaan media
sosial dan prestasi akademik mahasiswa menggunakan metode Simple Linear Regression. Data
penelitian diperoleh melalui penyebaran kuesioner kepada 41 mahasiswa sebagai responden.
Variabel independen dalam penelitian ini adalah durasi penggunaan media sosial per hari,
sedangkan variabel dependen adalah Indeks Prestasi Kumulatif (IPK) mahasiswa. Hasil analisis
menghasilkan persamaan regresi Y = 3,4466 − 0,0275X, yang menunjukkan arah hubungan negatif antara durasi penggunaan media sosial dan prestasi akademik mahasiswa. Nilai koefisien korelasi
sebesar r = -0,1527 mengindikasikan hubungan negatif yang sangat lemah, sedangkan nilai
koefisien determinasi sebesar R² = 0,0233 menunjukkan bahwa durasi penggunaan media sosial
hanya mampu menjelaskan 2,33% variasi prestasi akademik mahasiswa. Temuan ini menunjukkan
bahwa durasi penggunaan media sosial memiliki keterkaitan yang lemah dengan prestasi akademik
mahasiswa, sementara sebagian besar variasi prestasi akademik dipengaruhi oleh faktor lain di luar
penelitian. Dengan demikian, durasi penggunaan media sosial bukan merupakan faktor dominan
yang menentukan prestasi akademik mahasiswa.

Analisis Perbandingan Algoritma Apriori dan FP-Growth dalam Asosiasi Data Mining pada Data Transaksi Minimarket FS Barokah Berbasis Google Colab

Bagus Aris Setiawan — 2026-08-15

Data Mining is a data analysis technique used to discover hidden patterns in transaction data to support business decision-making. This study aims to analyze consumer purchasing patterns and compare the performance of the Apriori and Frequent Pattern Growth (FP-Growth) algorithms using transaction data from FS Barokah Minimarket. The dataset consisted of 1,000 transactions with 27 unique product items and underwent data cleaning, transformation, and one-hot encoding processes. The analysis was conducted using Google Colab with the pandas and mlxtend libraries. The results showed that both Apriori and FP-Growth produced identical outputs, namely 139 frequent itemsets and 224 association rules. The strongest association pattern was found between Air Galon and Chocolate with a lift ratio of 1.31, followed by Biscuits and Cheese with a lift ratio of 1.24. In terms of execution time, Apriori performed faster than FP-Growth on the analyzed dataset. These findings can support cross-selling strategies, product bundling, and more effective product placement in FS Barokah Minimarket.