PRE-PROCESSING DATA ON MULTICLASS CLASSIFICATION OF ANEMIA AND IRON DEFICIENCY WITH THE XGBOOST METHOD

Abstract

Anemia and iron deficiency are health problems in Indonesia and globally. In Multiclass Classification, data problems often occur, such as missing data, too many variables, and unbalanced data. Then pre-processing data will be carried out using MissForest imputation, Boruta featuring selection, and SMOTE to help improve the performance of the classification model in predicting a particular class. After the data pre-processing process is carried out, classification modeling will be carried out using the XGBoost algorithm. It was found that when pre-processing the data could improve the performance of the model in predicting multiclass classification for cases of anemia and iron deficiency in women in Indonesia by 0.815 for the accuracy value and 0.9693 for the AUC value.