Bayesian variable selection in modelling geographical heterogeneity in malaria transmission from sparse data: an application to Nouna Health and Demographic Surveillance System (HDSS) data, Burkina Faso

Table 2 Lag times and predictors selected by the variable selection

Model	Zero-inflated binomial	Zero-inflated binomial	Zero-inflated negative binomial	Zero-inflated negative binomial
Binomial component	*An. funestus*	*An. gambiae*	*An. funestus*	*An. gambiae*
Parameter
Rainfall	Lag 3	Lag 0	Lag 0	Lag 5
Vegetation (NDVI)	-	-	Lag 0	-
Day temp (LSTD)	Lag 4	Lag 3	Lag 1	-
Night temp (LSTN)	-	Lag 5	-	Lag 4
Distance to water body	Yes	No	No	No
Mixing Proportion
Rainfall	-	-	-	-
Vegetation (NDVI)	-	-	-	-
Day temp (LSTD)	-	-	Lag 1	Lag 2
Night temp (LSTN)	-	-	Lag 1	Lag 5
Distance to water body	Yes	Yes	Yes	No

Legend:
Lag 0: Average over the current month.
Lag 1: Average of the environmental covariate over the previous month.
Lag 2: Average of the environmental covariate over second previous month.
Lag 3: Average of the environmental covariate over the current and the previous month.
Lag 4: Average of the environmental covariate over the previous and the second previous month.
Lag 5: Average of the environmental covariate over the current, previous and the second previous month.

ISSN: 1756-3305