unable to fit the shared multiscale model for the count data using two datas sets(one is of district level and another state level)
Pooja B S
#joint modelling
#importing district level data on crime and corresponding shape file of the district
NE <- read_excel("D:/New_Research/NE_data.xlsx")
crime.shapeNE <- readOGR(dsn="D:/New_research", layer="NE")
merged <- merge(crime.shapeNE, NE, by='censuscode')
nb=poly2nb(crime.shapeNE)# neighbourhood matrix
head(nb)
nb2INLA("map.adj", nb)
g <- inla.read.graph(filename = "map.adj")
merged$re_u <- 1:nrow(merged@data)
#preparing state level data on crime and importing the shape file of the state
k=crime.shapeNE_states <- readOGR(dsn="D:/New_research", layer="NE_State")
ST_CD=c(18,12,13,14,15,16,17)
FemPop=c(17011147, 736799, 1049838, 1560229, 595855, 1979496, 1623063)
crime1.num=c(26352, 281, 39, 247, 172, 874, 568)
NE_state=data.frame(ST_CD, FemPop, crime1.num)
R1=sum(NE_state$crime1.num)/sum(NE_state$FemPop)
NE_state$Ecrime1=R1*NE_state$FemPop
merged1 <- merge(k, NE_state, by='ST_CD')
#preparing data for joint modelling
m=nrow(merged1@data)
s=nrow(merged@data)
l=m+s
Y=Matrix(NA, ncol = 2, nrow = l)
Y1=merged1$crime1.num
Y2=merged$crime.num.y
Y[1:m,1]=Y1
Y[8:l,2]=Y2
#intercept
Inter=matrix(NA, nrow = l, ncol = 2)
Inter[1:m,1]=1
Inter[8:l,2]=1
#offset
E1=merged1$Ecrime1
E2=merged$Ecrime
E=c(E1, E2)
ID.shared1=c(as.numeric(merged1$ST_CD),rep(NA, s))
ID.shared2=c(rep(NA, m),as.numeric(merged$ST_CEN_CD.x))
ID2=c(rep(NA,7), merged$re_u)
# model
form1=Y~0+Inter+ f(ID.shared1, model="bym", graph=g1) + f(ID.shared2, copy= 'ID.shared1', model="bym",graph=g1) + f(ID2, model="bym", graph=g)
fit=inla(form1, family = c("poisson", "poisson"), data = list(Y=Y, Inter=Inter, E=E, ID.shared1=ID.shared1, ID.shared2=ID.shared2, ID2=ID2), E=E, control.compute = list(waic=T), control.predictor = list(compute = TRUE))
what is the mistake I have done here.
Error in inla.core(formula = formula, family = family, contrasts = contrasts, :
f(ID.shared1). Covariate does not match 'values' 7 times. Indexes for mismatch:1 2 3 4 5 6 7. This is not what you want. Use NA values in the covariate!
*** inla.core.safe: inla.program has crashed: rerun to get better initial values. try=1/2
Error in inla.core(formula = formula, family = family, contrasts = contrasts, :
f(ID.shared1). Covariate does not match 'values' 7 times. Indexes for mismatch:1 2 3 4 5 6 7. This is not what you want. Use NA values in the covariate!
*** inla.core.safe: inla.program has crashed: rerun to get better initial values. try=2/2
Error in inla.core(formula = formula, family = family, contrasts = contrasts, :
f(ID.shared1). Covariate does not match 'values' 7 times. Indexes for mismatch:1 2 3 4 5 6 7. This is not what you want. Use NA values in the covariate!
Error in inla.core.safe(formula = formula, family = family, contrasts = contrasts, :
*** Fail to get good enough initial values. Maybe it is due to something else.
Helpdesk (Haavard Rue)
This seems straight forward. and its look like its something minor only.
btw, you use 'g1' but I don't see it.
In
f(ID.shared2, copy= 'ID.shared1', model="bym",graph=g1)
f(ID.shared2, copy= 'ID.shared1')
as it take all other infor from 'ID.shared1'
make sure 'ID.shared1' are integers in the range 1...7
If you cannot make it work, send me data/ complete code
H
> You received this message because you are subscribed to the Google
> Groups "R-inla discussion group" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to r-inla-discussion...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/r-inla-discussion-group/a362949a-f649-4227-9c5b-a5f2a98837a6n%40googlegroups.com
> .
--
Håvard Rue
he...@r-inla.org