# code adapted from MW but preparing for rjags instead of BUGS

# Options: numpc (set in calling file)
# numpc <- 10 or numpc <- 4

# Read file with data filenames
load("data_allproxy1209.R")
load("data_instrument.R")
setwd("bayesmodel")

# Parameters:
startdate <- 1000
numlags <- 2
numsim <- 100

# Get proxies going back to start date:
sel <- allproxy1209info[,"StartYear"] <= startdate
tmp <- colnames(allproxy1209)[-1][sel]
proxy <- allproxy1209[,tmp]
sel <- rowSums(is.na(proxy))==0
proxy <- proxy[sel,]
years <- allproxy1209[sel,1]
rownames(proxy) <- years
sel <- instrument[,"Year"] %in% years
temp <- instrument[sel,-1]
rownames(temp) <- instrument[sel,"Year"]

#adjust proxy selection to remove unreplicated tree ring data screened out by Mann et al 2008

proxy <- proxy[,-c(87:88)]      # to match method in earlier figures
proxy <- proxy[,-c(1,2,3,4,7,10,16,17,18,19,23,25,26,28,29,30,40,41,42,47,48,49,50,56,58,61,63,64,65,67,72,73,74,75,78,91)]

#remove tiljander
proxy <- proxy[,-c(51,52)]

# Run proxy pca:
mypc <- prcomp(proxy, center=TRUE, scale=TRUE)

# Prepare data for regressions:
in_idxI <- rownames(temp)
in_idxD <- as.character( sort(as.numeric(in_idxI), decreasing=TRUE) )
out_idxI <- setdiff(rownames(mypc$x), in_idxI)
out_idxD <- as.character( sort(as.numeric(out_idxI), decreasing=TRUE) )
y_in <- temp[in_idxD,1]
tmp <- matrix(NA, length(y_in), numlags)
rownames(tmp) <- in_idxD
colnames(tmp) <- paste("Lag",1:numlags,"_y",sep="")
for(i in 1:numlags){
  tmp[(1+i):length(y_in),i] <- y_in[1:(length(y_in)-i)]
}
x_in <- data.frame(tmp, mypc$x[in_idxD,1:numpc])
sel <- is.na(x_in)

tmp <- max( as.numeric(rownames(x_in))[rowSums(sel)>0] )
rownames(instrument) <- as.character(instrument[,1])
x_in[1,1] <- instrument[as.character(tmp+1),2]
x_in[1,2] <- instrument[as.character(tmp+2),2]
x_in[2,2] <- instrument[as.character(tmp+1),2]

# Fit linear regression:
mymodel <- lm(y_in~., data=x_in)
x_out0 <- data.frame(matrix(NA, length(out_idxD), numlags), mypc$x[out_idxD,1:numpc])
colnames(x_out0) <- colnames(x_in)
x_out <- data.frame(rbind(x_in,x_out0))

# Prep for rjags:
mydata <- cbind(y_in, x_in)
colnames(mydata) <- c("y[]", "Lagy[,1]", "Lagy[,2]", paste("PC[,", 1:numpc, "]", sep=""))
write.table(mydata, paste("rjags_data","_",numpc,".txt", sep=""), sep="\t", row.names=FALSE, quote=F)

setwd("../")