A New Body Weight Lifelog Outliers Generation Method: Reflecting Characteristics of Body Weight Data

Lifelogs are generated in our daily lives and contain useful information for health monitoring. Nowadays, one can easily obtain various lifelogs from a wearable device such as smartwatch. These could include noise outliers. In general, the amount of outliers is significantly smaller than that normal data, resulting class imbalance. To achieve good analytic accuracy, should be filtered. have specific characteristics: low volatility periodicity. It very important to continuously analyze manage them within time. solve imbalance problem weight lifelog we propose new outlier generation method reflects characteristics body weight. This study compared proposed with SMOTE-based data augmentation GAN-based methods. Our results confirm detection was better SVM, XGBOOST, CATBOOST algorithms. Through them, reduce level, improve quality, analytics accuracy.