# Authors: Fuyu Frank Xu and Kate Beard, School of Computing and Information Science, University of Maine, Orono, Maine
# Function name: STES.sim3().
# This function is used after running STES.sim1 and outputs not only global and local similarity matrices separately but also an ingetrated matrix including
# pairwise global and local similarity values.
#
# Modified date: May, 2019
STES.sim2 <- function(data,ctw=1){
# ----------------------------------------------- Test for errors if input data is not matrix format ------------------------------
#
if(!is.matrix(data)){
print("!--------------------- ERROR #001 -----------------!")
print("! Input data file 'data' is not a matrix! !")
print("!----------------------------------------------------!")
return()
} # end of matrix input check
# ----------------------------------------------- Eventization --------------------------------------------------------------------
nr = nrow(data) #number of rows
nc = ncol(data) #number of columns
nw = nc%/%ctw #number of complete local temporal windows
nm = nc%%ctw #modulus
cnp = nw #column number of pairwise similarity matrix
if (nm != 0) { cnp = cnp + 1}
#calculate number of rows for the matrix of pairwise similarity
#the same number of pairwise comparisons
rnp = 0 #initialize row number of pairwise comparisons
for (i in 1:(nr-1)){
for (j in (i+1):nr){
rnp = rnp + 1
}
}
#create an empty matrix for storing global and local pairwise similarity
grandsim <- matrix( , rnp, cnp+1)
rnames <- rep(NA, rnp) #store new row names for combined similarity matrix
cnames <- rep(NA, cnp+1) #store new column names for combined similarity matrix
cat('Original STES matrix', "\n")
print(data)
cat("\n")
globalsim = STES.sim1(data=data)
rnp = 0
for (i in 1:(nr-1)){
for (j in (i+1):nr){
rnp = rnp + 1
grandsim[rnp,1] = globalsim[i,j]
rnames[rnp] = paste(rownames(globalsim)[i], rownames(globalsim)[j], sep = "-")
}
}
cnames[1] = "Entire" #column for global similarity
cat('Global event sequence similarity matrix', "\n")
print(globalsim)
cat("\n")
for (i in 1:nw){
subdata = data[,((i-1)*ctw+1):(i*ctw)]
localsim = STES.sim1(data=subdata)
#combine local similarity matrix to an integrated grand matrix
cnames[i+1] = paste("Window",i, sep = "-")
rnp = 0
for (j in 1:(nr-1)){
for (k in (j+1):nr){
rnp = rnp + 1
grandsim[rnp,i+1] = localsim[j,k]
}
}
cat('Local similarity matrix at window ', i, "\n")
print(localsim)
}
if (nm != 0){
subdata = data[,(nw*ctw+1):(nw*ctw+nm)]
localsim = STES.sim1(data=subdata)
cnames[cnp+1] = paste("Window",cnp, sep = "-")
rnp = 0
for (j in 1:(nr-1)){
for (k in (j+1):nr){
rnp = rnp + 1
grandsim[rnp,cnp+1] = localsim[j,k]
}
}
cat('Local similarity matrix at last incomplete window ', nw+1, "\n")
print(localsim)
}
rownames(grandsim) = c(rnames)
colnames(grandsim) = c(cnames)
cat("\n")
cat("Integrated global and local similarity matrix", "\n")
print(grandsim)
} # end function STES.sim3