Predicting household per capita expenditure by using log-normal regression in Poso Regency Central Sulawesi

Nila Ayu Nur Roosyidah*, Achmad Choiruddin, Ismaini Zain

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

Abstract

Predicting household per capita expenditure accurately is necessary to measure social welfare of such region, for example its poverty level. In Indonesia, as well as Poso Regency, that variable tends to be highly skewed and heteroscedastic. Log-transformation is usually considered to handle those issue in the standard linear regression. However, that method only works well on minor-skewed data and can't measure the absolute effect of each correspondence factor. In this study, we propose log-normal regression in tackling this issue. The results show that compared to the traditional method, log-normal regression performs much better and predicts more accurately.

Original languageEnglish
Pages (from-to)512-519
Number of pages8
JournalProcedia Computer Science
Volume234
DOIs
Publication statusPublished - 2024
Event7th Information Systems International Conference, ISICO 2023 - Washington, United States
Duration: 26 Jul 202328 Jul 2023

Keywords

  • highly-skewed data
  • household per capita expenditure
  • poverty
  • social transfer
  • standard linear regression

Fingerprint

Dive into the research topics of 'Predicting household per capita expenditure by using log-normal regression in Poso Regency Central Sulawesi'. Together they form a unique fingerprint.

Cite this