Assessing dengue forecasting methods: a comparative study of statistical models and machine learning techniques in Rio de Janeiro, Brazil

Table 4 Accuracy of the machine learning techniques considered with covariates temperature and humidity at various forecast horizons

Week	SVM			Random Forest			XGBoost			LSTM			Prophet
Week	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE
1	112.29	29.84	176.15	410.44	119.94	621.41	439.89	151.00	644.90	71.35	22.31	101.53	200.20	46.62	293.64
2	138.34	36.41	215.68	418.29	119.06	628.82	440.81	139.93	646.74	89.50	23.56	130.71	222.88	49.96	326.45
3	165.67	46.73	248.43	416.61	80.38	628.41	431.51	93.65	636.03	114.73	26.24	173.90	247.76	54.57	353.17
4	193.32	57.77	280.87	425.16	79.53	635.42	452.01	92.57	653.74	148.74	32.14	223.35	265.58	57.98	373.31
8	289.62	108.35	405.34	444.25	69.99	652.34	461.87	78.97	660.34	292.91	57.01	416.62	316.96	64.81	444.14
12	360.53	175.58	492.24	463.77	69.96	667.65	465.48	73.89	665.71	410.96	79.92	573.32	342.47	61.86	495.31

Week	SVM-Lag			Random Forest-Lag			XGBoost-Lag			LSTM-Lag			Prophet-Lag
Week	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE	MAE	MAPE (%)	RMSE
1	188.99	50.10	310.12	414.14	143.38	603.79	412.62	172.51	610.86	83.58	28.87	117.39	208.55	77.77	299.80
2	205.94	56.69	338.36	422.68	142.14	613.06	433.29	210.35	627.08	102.1	28.27	155.82	235.76	65.11	339.18
3	223.76	63.06	364.19	424.42	117.13	616.01	416.41	145.61	611.33	132.8	30.88	201.65	259.27	58.71	369.47
4	245.03	70.38	391.97	428.85	109.80	621.38	427.03	154.94	612.99	166.45	34.97	251.28	280.24	59.53	394.74
8	303.25	85.87	470.16	439.87	71.77	641.60	431.1	74.68	630.37	304.93	60.16	438.11	347.77	70.88	479.95
12	341.82	121.50	513.45	454.75	68.21	655.17	446.64	71.04	646.80	422.09	82.63	586.77	389.37	74.56	540.32

ISSN: 1349-4147