Mid-Epidemic Forecasts of COVID-19 Cases and Deaths: A Bivariate Model Applied to the UK

Table 3

Criteria for out-sample predictions (M is number of days in training sample; F = 20 in all cases).

Criterion	Poisson-gamma (PG)				Poisson-lognormal (PLN)				Poisson-log-Student (PLS)
	Mean	2.5%	Median	97.5%	Mean	2.5%	Median	97.5%	Mean	2.5%	Median	97.5%
M = 80
Average daily cases	4357	1818	3686	7407	2214	1078	1881	4875	3118	1043	2869	5652
Average daily deaths	609	344	603	904	613	303.8	613.4	921.8	587	0	514	1471
ωc	0.41				0.03				0.16
ωd	0.34				0.38				0.40
Actual daily averages in period (M + 1, M + F)
New cases	4857
New deaths	662
M = 100
Average daily cases	2346	1523	2353	3173	1975	1384	1947	2658	2630	622	2565	4340
Average daily deaths	265	184	265	351	224.8	135.8	221.5	332.8	275	52	276	510
ωc	0.09				0.00				0.31
ωd	0.06				0.03				0.32
Actual averages in period (M + 1, M + F)
Cases	3039
Deaths	330
M = 120
Average daily cases	1132	1006	1129	1274	1133	963.2	1136	1320	1310	604	1358	1835
Average daily deaths	129	89	129	170	117.9	77.6	117.6	160.4	135	37	134	236
ωc	0.00				0.00				0.42
ωd	0.00				0.00				0.07
Actual averages in period (M + 1, M + F)
Cases	1506
Deaths	216