Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal JOURNAL OF INFORMATION SYSTEM RESEARCH (JOSH) Jurnal Teknik Informatika (JUTIF)
Faran, Jhiro
Unknown Affiliation
Computer Science & IT Decision Sciences, Operations Research & Management

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

Found 2 Documents
Search

 1 
ANALYSIS OF THE EFFECTIVENESS OF POLYNOMIAL FIT SMOTE MESH ON IMBALANCE DATASET FOR BANK CUSTOMER CHURN PREDICTION WITH XGBOOST AND BAYESIAN OPTIMIZATION Faran, Jhiro; Triayudi, Agung
Jurnal Teknik Informatika (Jutif) Vol. 5 No. 3 (2024): JUTIF Volume 5, Number 3, June 2024
Publisher : Informatika, Universitas Jenderal Soedirman

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52436/1.jutif.2024.5.3.1284

Abstract

The case of churn in the banking industry, namely customers who leave or no longer use bank services, is a serious problem that requires an appropriate solution. The aim of this research is to predict churn and take appropriate preventive actions using machine learning. The dataset contains 10,000 bank customer data with 14 relevant features. Only about 20% of customers experience churn, creating a data imbalance problem in classification. To overcome data imbalances, the SMOTE oversampling technique was applied. Also introduced was the development of the SMOTE technique, namely, Polynomial Fit SMOTE Mesh (PFSM). PFSM works by combining each point in the data with a linear function and producing synthetic data at each connected distance. Experimental results show that the model developed using PFSM and optimized with Bayesian Optimization for the XGBoost algorithm achieved 86.1% accuracy, 70.87% precision, 53.81% recall, and 61.17% F-score. This indicates that the approach is successful in improving predictive capabilities and identifying potential customers for churn earlier. This research has significant relevance in the banking industry, helping banks to safeguard their customers and improve banking business performance..
Perbandingan Algoritma K-Means dan K-Medoids Dalam Pengelompokan Kelas Untuk Mahasiswa Baru Program Magister Faran, Jhiro; Aldisa, Rima Tamara
Journal of Information System Research (JOSH) Vol 5 No 2 (2024): Januari 2024
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/josh.v5i2.4753

Abstract

This research discusses a comparison of two grouping algorithms, namely K-Means and K-Medoids, in the context of class grouping for new master's program students. Choosing the right clustering algorithm can help universities optimize resource allocation and maximize student learning experiences. K-Means is a popular clustering algorithm, which works by dividing data into a number of homogeneous groups based on the distance between data points and the cluster center. Meanwhile, K-Medoids is a variation of K-Means that uses actual data points as a cluster representation, which makes it more resistant to outliers. This research involves a dataset of new master's program students which includes various attributes, such as entrance exam scores, educational background, and major preferences. The comparison results between K-Means and K-Medoids were carried out by considering clustering evaluation metrics such as SSE (Sum of Squared Errors) and Silhouette Score. Experimental results show that the performance of K-Means and K-Medoids differs depending on the characteristics of the dataset. K-Means tends to produce more homogeneous groups, but is more sensitive to outliers. In contrast, K-Medoids tend to be more stable in dealing with outliers, but may produce less homogeneous groups. Therefore, the selection of an appropriate clustering algorithm should be based on the specific goals and characteristics of the new master's program student population. This research provides valuable insight for colleges in planning the allocation of classes, mentors, and other resources for new students. The right decisions in class grouping can increase student retention, learning satisfaction, and academic success. In addition, this research also stimulates further discussion in combining different clustering methods to achieve more optimal results in grouping classes of new master's program students.
Co-Authors Agung Triayudi Aldisa, Rima Tamara