p-Index From 2021 - 2026
5.446
P-Index
Claim Missing Document
Check
Articles

Found 3 Documents
Search
Journal : Kubik

Penerapan Extreme Learning Machine Dalam Meramalkan Harga Minyak Sawit Mentah Siti Aisyah; Nurissaidah Ulinnuha; Abdulloh Hamid
KUBIK Vol 7, No 2 (2022): KUBIK: Jurnal Publikasi Ilmiah Matematika
Publisher : Jurusan Matematika, Fakultas Sains dan Teknologi, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/kubik.v7i2.20460

Abstract

The need for crude palm oil has increased due to the large demand for vegetable oils in various parts of the world. Beginning in March 2022, the price of crude palm oil set a record high which caused international cooking oil prices to soar, especially for Indonesia. This study aims to predict the price of crude palm oil with test parameters, namely hidden neurons and activation functions. The method used is Extreme Learning Machine (ELM). This method is a development of the artificial neural network (ANN) method which can overcome weaknesses in the learning speed process. There are several stages in this study: (1) pre-processing the data by normalizing the data and dividing the data using the time series split method, (2) analyzing the data using the ELM method by testing parameters, namely hidden neurons and activation functions, (3) analyzing the results of the best parameter trials, (4) calculating forecasting data using the best parameters that have been obtained, and (5) analyzing the forecasting results that have been obtained. This study uses daily data on the price of crude palm oil from April 1 2021 to April 14 2022 obtained from the Investing website. The results of the research that has been carried out obtained MAPE and RMSE values of 0.0173 and 0.0308 with the best parameters namely the number of hidden neurons of 5 and the binary sigmoid activation function. Based on the results obtained, it is hoped that it will make it easier for the government to determine the price of crude palm oil in the future.
Klusterisasi Penyandang Masalah Kesejahteraan Sosial (PMKS) Di Kabupaten Bojonegoro Menggunakan Algoritma K-Medoids Elisa Syafaqoh; Nurissaidah Ulinnuha; Lutfi Hakim
KUBIK Vol 7, No 2 (2022): KUBIK: Jurnal Publikasi Ilmiah Matematika
Publisher : Jurusan Matematika, Fakultas Sains dan Teknologi, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/kubik.v7i2.21653

Abstract

Persons with Social Welfare Problems (PMKS) are individuals, community groups, or families who cannot adequately and properly meet their economic, physical, mental, and social needs, both spiritually and physically, because of an obstacle, difficulty, or disturbance. This study aimed to classify sub-districts in Bojonegoro Regency based on the level of social welfare problems using the K-Medoids Clustering (PAM) Analysis method. There are 2 clusters formed with an Average Silhouette of 0.73. Cluster 1 is a sub-district group with common social welfare problems, and Cluster 2 is a sub-district group with high social welfare problems. Each silhouette value of the cluster is 0.74 and 0.70 with the specifications of a well-formed and strong structure.
Implementation of BiLSTM to Predict World Crude Oil Prices Sari, Firda Yunita; Ulinnuha, Nurissaidah
KUBIK Vol 10 No 1 (2025): IN PRESS
Publisher : Jurusan Matematika, Fakultas Sains dan Teknologi, UIN Sunan Gunung Djati Bandung

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

Abstract

The main source of energy worldwide is crude oil, which is used by almost all countries as an energy source. Crude oil plays a key role in driving the global economy, especially in the industrial and transportation sectors. Along with technological developments, crude oil price predictions can be made more sophisticated using artificial intelligence-based methods, one of which is the Bidirectional Long Short-Term Memory (BiLSTM) method which is a development of the Long Short-Term Memory (LSTM) method by combining past and future information when processing sequential data, BiLSTM uses forward and backward LSTM simultaneously to increase accuracy. The study used world crude oil price data for 1 year. There are 57 tests with several parameters such as data division, number of neurons, batch size, and activation function. After testing with the BiLSTM method for 57 scenarios, there is the smallest MAPE value of 0.09% at a data division of 90:10, number of neurons 100, batch size of value 4, and ReLu activation function. The resulting prediction model is highly accurate based on the MAPE criterion value.