Predicting the incidence of infectious diarrhea with symptom surveillance data using a stacking-based ensembled model

Table 3 Prediction performance of stacking ensembled model

Selection of base models	Model names	RMSE	MAE	MAPE
The base models with the lowest RMSE	ANN	86.57	60.47	17.38%
	LSTM	84.85	59.07	16.99%
	SVR	86.41	61.31	16.37%
	XGBoost	85.13	65.75	19.16%
	Stacking	74.38	55.22	16.01%
The base models with the lowest MAE	ANN	86.57	60.47	17.38%
	LSTM	88.67	57.50	16.52%
	SVR	86.41	61.31	16.37%
	XGBoost	93.43	63.12	17.31%
	Stacking	77.70	58.54	16.88%
The base models with the lowest MAPE	ANN	92.75	63.13	16.83%
	LSTM	90.90	59.07	15.92%
	SVR	86.41	61.31	16.37%
	XGBoost	88.00	63.61	17.25%
	Stacking	75.82	55.93	15.70%

Note The underlined model was the best stacking ensembled model according to MAPE

ISSN: 1471-2334