Record Details

Pengaruh Jumlah Record Dataset Terhadap Algoritma Klasifikasi Berdasarkan Data Customer Churn

Jurnal Ilmiah Informatika

View Archive Info
 
 
Field Value
 
Title Pengaruh Jumlah Record Dataset Terhadap Algoritma Klasifikasi Berdasarkan Data Customer Churn
 
Creator Anita
Wicaksono, Anggito
Padilah, Tesa Nur
 
Subject Dataset
Customer Churn
Classification
Data Mining
CRISP-DM
 
Description Telecommunication is one of the fastest growing industrial sectors so that there are more telecommunication companies. This can create various threats if the company does not use the strategy properly. Customer churn refers to the level of customer reduction which is one of the threats to reducing the company's revenue. This is an important issue for developing companies to evaluate in order to reduce the potential for churn that occurs. The initial stage that needs to be done is to predict customers who have the potential to switch from the company, one of which is the data mining approach. Classification is a data mining technique that can predict the class of datasets with various existing classification algorithms. The purpose of this study is to identify the effect of the number of dataset records on several classification algorithms. This research was conducted based on the CRISP-DM method by applying three classification algorithms, namely Logistic Regression, Naïve Bayes, and Decision Tree C4.5. The results showed that the greater the number of records in the dataset, the higher the accuracy value will be obtained. In dataset-1, logistic regression is a better algorithm based on an accuracy value of 80.09%, while naïve Bayes is superior based on an AUC value of 0.733 and an execution time of 0.00798 seconds. In dataset-2, it is found that decision tree is an algorithm that is more suitable than logistic regression and naïve Bayes algorithms, with an accuracy of 91.9% and an AUC value of 0.846 which is included in the good classification criteria. However, in execution time, the naïve Bayes algorithm only takes a processing time of 0.00403 seconds.
 
Publisher Department of Science and Technology Ibrahimy University
 
Date 2021-06-29
 
Type info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
Peer-reviewed Article
 
Format application/pdf
 
Identifier https://journal.ibrahimy.ac.id/index.php/JIMI/article/view/1223
10.35316/jimi.v6i1.1223
 
Source Jurnal Ilmiah Informatika; Vol. 6 No. 1 (2021): Jurnal Ilmiah Informatika; 1-10
2549-6301
2549-7480
10.35316/jimi.v6i1
 
Language eng
 
Relation https://journal.ibrahimy.ac.id/index.php/JIMI/article/view/1223/974
 
Rights Copyright (c) 2021 Jurnal Ilmiah Informatika
https://creativecommons.org/licenses/by-nc/4.0