Skip to main navigation Skip to search Skip to main content

랜덤포레스트를 이용한 셀룰로오스 아세테이트의 치환도 예측 모델링과 변수 중요도 분석

Translated title of the contribution: Feature Importance Measures from Random Forest Regressor for Predicting Degree of Substitution in Cellulose Acetate
  • Yong Ju Lee
  • , Ji Eun Cha
  • , Geon Woo Kim
  • , Tai Ju Lee
  • , Jieun Lee
  • , Sang Jin Chun
  • , Jaegyoung Gwon
  • , Hyoung Jin Kim*
  • *Corresponding author for this work
  • Kookmin University

Research output: Contribution to journal › Journal article › peer-review

Abstract

The degree of substitution (DS) is the key parameter for characterizing cellulose acetate (CA). While DS is typically measured by titration, this method is time consuming, labor-intensive, and requires hazardous chemicals. To provide an alternative, a random forest (RF) regression model was developed to predict the DS of CA using infrared spectroscopy. By deacetylation process, CA samples were prepared with DS values ranging from 0.45 to 2.85. The RF model was trained on spectral data in the region of 1500–800 cm-1, selected for its high feature importance, achieving an R2 of 0.988, consistent with the full spectral range (4000–600 cm-1). These results demonstrate that the RF model offers a robust, interpretable, and efficient approach for DS prediction in CA.

Translated title of the contributionFeature Importance Measures from Random Forest Regressor for Predicting Degree of Substitution in Cellulose Acetate
Original languageKorean
Pages (from-to)26-39
Number of pages14
JournalPalpu Chongi Gisul/Journal of Korea Technical Association of the Pulp and Paper Industry
Volume56
Issue number6
DOIs
StatePublished - 2024
Externally publishedYes

Keywords

  • cellulose derivatives
  • DBSCAN
  • Infrared spectroscopy
  • multilayer perceptron (MLP)
  • partial least square regression (PLSR)

Fingerprint

Dive into the research topics of 'Feature Importance Measures from Random Forest Regressor for Predicting Degree of Substitution in Cellulose Acetate'. Together they form a unique fingerprint.

Cite this