Jurnal Statistika dan Komputasi

Pengelompokkan Wilayah Banjir di Jawa Tengah untuk Mitigasi Banjir Menggunakan Pendekatan K-Medoids

2024-11-23T10:06:10+00:00

Background: Flooding is a natural event that can occur at any time, often resulting in fatalities and significant material losses. Mapping flood zones in Central Java based on flood occurrences is crucial for optimizing disaster management. Clustering approaches are highly relevant and potential methods for tackling flood mitigation challenges in Central Java.

Objective: To map flood zoning in Central Java using the optimal K-Medoids method based on the Silhouette Coefficient.

Methods: This study uses the K-Medoids method for Clustering analysis because it is more resistant to outliers. Unlike K-Means, K-Medoids selects the medoid as the cluster center, making it more stable against extreme values. The data used was obtained from the Dinas PUSDATARU of Central Java Province regarding flood events in the region from October 1, 2022, to March 2023.

Results: The K-Medoids method with k=2 produced the highest Silhouette Coefficient of 0.83748, Clustering 34 districts/cities with low flood occurrences and 1 district/city with high flood occurrences. This model evaluation supports the planning of disaster mitigation policies that focus more on flood-prone areas.

Conclusion: There are two groups of districts/cities based on flood occurrence levels. The high Silhouette Coefficient value indicates a good cluster structure.

Implementasi Metode Seasonal Autoregressive Integrated Moving Average (SARIMA) untuk Memprediksi Curah Hujan di Kota Semarang

2024-12-03T06:07:28+00:00

Background: Rainfall is one of the important factors that has a significant impact on various aspects of life, especially in urban areas such as Semarang. Significant fluctuations in rainfall can cause flooding, which negatively impacts infrastructure, agriculture, health and well-being of the community. Therefore, accurate rainfall forecasting is essential to support informed decision-making.

Objective: The purpose of this study is to identify and build an optimal SARIMA model for rainfall forecasting in Semarang City.

Methods: This study used the Seasonal Autoregressive Integrated Moving Average (SARIMA) method to analyze the monthly rainfall data of Semarang City for the period 2017-2022, because it was able to handle seasonal patterns in the time series data. The best model is determined based on the Akaike Information Criterion (AIC) value, while the accuracy of the prediction is measured using the Mean Absolute Percentage Error (MAPE) value.

Results: Based on the results of the analysis, the best SARIMA model was SARIMA (1,1,0) (0,1,0)12 because it produced the smallest AIC value (121.67) and MAPE of 41.59%. This model is used to predict rainfall from January 2023 to December 2025.

Conclusion: The SARIMA (1,1,0) (0,1,0)¹² model is the best model for rainfall forecasting in Semarang City. The results of this study support previous studies that state that the SARIMA method is effective for rainfall data that have high fluctuations and extreme values.

Optimalisasi Peramalan Total Aset PT. BPD Kaltim Kaltara dengan Double Exponential Smoothing Brown

2024-12-03T14:06:10+00:00

Background: Total assets can provide a comprehensive picture of the wealth owned by a company or institution, with total assets also helping to assess the scale of operations, stability, and the company’s ability to meet its financial responsibilities. Study on the total assets held by PT. BPD Kaltim Kaltara is interesting to do because it has an important role in advancing economic growth in the East Kalimantan and North Kalimantan regions. Digital transformation can influence how assets grow and how capital is structured.

Objective: Predicting PT BPD Kaltim Kaltara’s total assets over the next three periods using the DES Brown method with the optimal constant.

Methods: Double Exponential Smoothing Brown (DES Brown) with constants α = β = 0.3; 0.6; 0.7; 0.8.

Results: The smallest MAPE value is obtained at the constant α = β = 0.3, indicating that the DES Brown method with this constant provides the most accurate forecasting results.

Conclusion: The forecasting results for the next three periods show a stable upward trend, namely September at Rp48,389,055.93, October at Rp48,480,301.62, and November at Rp48,571,547.30. Thus, the DES Brown method has proven effective in forecasting the total assets of PT. BPD Kaltim Kaltara and can be used to support the company's financial decision making.

Comparison of K-Means and Fuzzy C-Means for Optimizing Tuberculosis Management and Healthcare Service Allocation in Bojonegoro

2024-12-07T03:58:13+00:00

Background: According to the 2022 publication by BPS (Statistics Bureau) of Bojonegoro Regency, there were 1,765 tuberculosis cases spread across all districts in Bojonegoro. This number is disproportionate to the availability of healthcare workers, which totaled only 1,261, comprising medical personnel, nurses, midwives, and pharmacists.

Objective: This study aims to cluster districts in Bojonegoro Regency based on tuberculosis cases and healthcare workforce data by comparing the K-Means and Fuzzy C-Means methods. The objective is to identify which districts require more attention and which are already in better condition.

Methods: The best clustering method was determined using the Sum of Squared Error (SSE) criterion. The data used in this study was sourced from the Statistics Bureau, containing information on tuberculosis cases and the number of healthcare workers in each district..

Results: The result shows that K-Means achieved a lower SSE (4704.031) compared to Fuzzy C-Means (4854.247), which divided the district into 4 clusters: low, medium, and high. By categorizing the districts into these clusters, the Bojonegoro government is expected to better target its interventions and resources. Moreover, the government can evaluate districts with high tuberculosis cases to implement specific strategies.

Conclusion: This study concludes that K-Means with 4 clusters is the most effective method for this type of clustering.

Clustering Study Of Hospitals In Bojonegoro Based On Health Workers With K-Means And K-Medoids Methods

2024-12-18T13:14:28+00:00

Background: Hospitals are institutions that provide inpatient care for the sick. In Bojonegoro, hospital services are considered adequate. However, a shortage of nurses often requires patients' families to assist with care.

Objective: This research aims to compare clustering methods to find the best method that can be applied to cluster hospitals based on the type of health workers.

Methods: This study uses two clustering methods, namely K-Means and K-Medoids Clustering, which are compared to determine the best method. The data source used is secondary data, which consists of the number of medical staff for each medical position, obtained from the Satu Data Bojonegoro website in 2020.

Results: The K-means method proved to be the best for grouping healthcare workforce data. Its average within-cluster distance value is -6.763, the closest to zero. The K-means method resulted in 4 clusters. These include cluster_0 (3 hospitals), cluster_1 (1 hospital), cluster_2 (1 hospital), and cluster_3 (5 hospitals).

Conclusion: The clustering results show that K-Means with 4 clusters is the best method, with Cluster_0 and Cluster_3 having below-average health workers, and Cluster_1 and Cluster_2 having above-average health workers, with Cluster_2 having the highest and Cluster_3 the lowest number of health workers in Bojonegoro.