Table 6.

The performance of Random Forest models trained on different groups of features for daily patient count prediction. The performance was compared using the Pearson correlation coefficient (ρ), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE) from two types of cross-validation experiments.

	Random 70/30 split			Leave One Day Out
Feature	ρ	MAE	RMSE	ρ	MAE	RMSE
personTime	0.91	12.87	16.38	0.93	12.30	16.08
speechFeat	0.07	41.04	47.99	0.22	37.67	44.57
coughFeat	0.62	28.04	34.17	0.69	25.29	32.70
isHoliday, dayType	0.86	14.61	19.53	0.88	14.85	21.34
isHoliday, dayType, personTime, speechFeat, coughFeat	0.93	11.96	15.10	0.95	9.32	12.95
isHoliday, dayType, personTime	0.94	9.52	12.58	0.95	8.91	13.62