# RAPFish - Rapid Assessment of Fishery Sustaintability

# # for local execution in R

#
# 2012-08-24 Divya Varkey Additional 25 anchors not plotted
# 2012-04-15 Divya Varkey Rapfish, MC and Leverage
# 2011-09-27 Divya Varkey rapfish code for 1 field at a time
# 2011-08-14 Ken Lockhart fine tune rotation
# 2011-08-13 Ken Lockhart radarplots (Nx3 array) functional
# 2011-08-12 Ken Lockhart RAPplot functional
# 2011-08-11 Ken Lockhart Integrate Divya's plots in mainline.
# 2011-08-07 Divya Varkey Radar plot
# 2011-08-02 Divya Varkey MDS Plot w Legend
# 2011-07-23 Ken Lockhart Created.
#

# RAPFish expects an input matrix of RAPFish scorecards defining a set of

# fisheries. The fisheries data is preceded by a some of anchors (to control
# the orientation of results sets generated by the multidimensional scaling
# algorythm that is the heart of RAPFish analysis.
#
# attributes ...->
# Good 0 0 0 0 ..... 0
# Bad 0 0 0 0 ..... 0
# anchor1 0 0 0 0 ..... 0
# ...
# anchorN
# fishery1
# ...
# fisheryN

# ----PREREQUISITES-----------------------------------------
# Load required libraries
library(grid)

# ----INPUT--------------------------------------------------
loaddata <- function(file.name){
# Load the data into a data.frame
fisheries = read.csv(file.name,row.names=1,check.names=TRUE)
print(colnames(fisheries))
print(rownames(fisheries))

return(fisheries)
}

# ----toPolar----------------------------------------------
toPolar <- function(x,y) {
# Polar coordinates from Cartesian coordinates.
radius= sqrt(x^2 + y^2);
theta = atan(y/x);
theta = theta*(180/pi) # convert radians to degrees
return( c(radius,theta) )
}

# ----toCartesian------------------------------------------
toCartesian <- function (radius, theta) {
# Cartesian coordinates from Polar coordinates.
x= radius * cos(theta)
 y= radius * sin(theta)
return( c(x,y) )
}

# ----rotate------------------------------------------------
rotate <- function (x,y,angle) {
# 2D rotation of a cartesian coordinate
theta= angle* (pi/180) # convert degrees to radians
x= (x*cos(theta)) - (y*sin(theta))
y= (x*sin(theta)) + (y*cos(theta))
return( c(x,y) )
}

# ----NORMALIZE----------------------------------------------
normalize <- function(dataset) {
# normalize & scale the data
good= dataset[1,]
bad = dataset[2,]

results= NULL
for (i in 1:nrow(dataset)) {
drow = dataset[i,]
distance = ifelse(good<bad, bad-drow, drow)
distance = t(distance)
score = ( distance / (abs(good-bad)) )*100
rownames(score) = rownames(drow)
results = rbind(results, score)
}

return(results)
}

# ----MULTIDIMENSIONAL SCALING-------------------------------
mdscale <- function(dataset) {
# use multidimensional scaling to compare the fisheries
#
disttbl= dist(dataset,"euclidean",diag=TRUE,upper=TRUE)
coords = cmdscale(disttbl,k=2)

# Flip & Scale - when needed.

#dx = coords['GOOD',1] - coords['BAD',1]
if (coords['GOOD',1] < coords['BAD',1]) {
coords[,1]= -coords[,1] # flip horizontal
}

#dy = coords['GOOD',2] - coords['BAD',2]

if (coords['UP',2] < coords['DOWN',2]) {
coords[,2]= -coords[,2] # flip vertical
}

# Rotate - when needed.

p1= toPolar( coords['GOOD',1], coords['GOOD',2] )
p2= toPolar( coords['BAD',1], coords['BAD',2] )
angle= -( mean(c(p1[2],p2[2])))
coords1= matrix(rotate(coords[,1],coords[,2],angle),,2)
rownames(coords1)= rownames(coords)
 x.min= min(coords1[,1])
x.max= max(coords1[,1])
y.min= min(coords1[,2])
y.max= max(coords1[,2])

sc_x=100/(x.max-x.min)
X_scores=coords1[,1]*sc_x
ad_x=abs(X_scores[2])
X_scores=X_scores+ad_x

sc_y=100/(y.max-y.min)
Y_scores=coords1[,2]*sc_y

coords2=cbind(X_scores,Y_scores)

return(coords2)
}

# ----RAPPLOT--------------------------------------------------

RAPplot1 <- function(coords2,num_fish,n_an) {

# plot the Fishery rankings

cols=rainbow(num_fish,start=0, end=.7)
compass = coords2[1:4,]
anchors = coords2[5:n_an,]
ap=n_an-25
anchorsplot=coords2[5:ap,]
fish = coords2[(n_an+1):nrow(coords2),]
ind=seq(1,num_fish)
fish=cbind(fish,ind)
sfish = fish[order(fish[,1]),]

x.min1= compass[2,1]
x.max1= compass[1,1]
y.min1= compass[4,2]
y.max1= compass[3,2]

plot(coords2,type="n",xlab="",ylab="",main="",
xlim=c(x.min1,x.max1+30),las=1)
lines(c(x.min1,x.max1),c(0,0),col="dark grey",lwd=1.5)
lines(c(50,50),c(y.min1,y.max1),col="dark grey",lwd=1.5)
#grid
lines(c(25,25),c(y.min1,y.max1),col = "lightgray", lty = "dotted")
lines(c(75,75),c(y.min1,y.max1),col = "lightgray", lty = "dotted")

lines(c(x.min1,x.max1),c(25,25),,col = "lightgray", lty = "dotted")

lines(c(x.min1,x.max1),c(-25,-25),,col = "lightgray", lty = "dotted")

points(compass,xlab="",ylab="",col="dark red",pch=rownames(compass))
points(anchorsplot,xlab="",ylab="",col="dark red",pch=19,cex=0.6)
points(sfish,xlab="",ylab="",col=cols,pch=19)

legend(x='right',rownames(sfish),col=cols,pch=19,cex=0.7,bty="n")
output=list()
output$scores=sfish[,1:2]
 output$ind=sfish[,3]
output$cols=cols
return(output)
}

RAPplot2 <- function(coords2,num_fish,n_an) {

# plot the Fishery rankings

cols=rainbow(num_fish,start=0, end=.7)
compass = coords2[1:4,]
anchors = coords2[5:n_an,]
ap=n_an-25
anchorsplot=coords2[5:ap,]
fish = coords2[(n_an+1):nrow(coords2),]
ind=seq(1,num_fish)
fish=cbind(fish,ind)
sfish = fish[order(fish[,1]),]

x.min1= compass[2,1]
x.max1= compass[1,1]
y.min1= compass[4,2]
y.max1= compass[3,2]

plot(coords2,type="n",xlab="",ylab="",main="",
xlim=c(x.min1,x.max1+50),las=1)
lines(c(x.min1,x.max1),c(0,0),col="dark grey",lwd=1.5)
lines(c(50,50),c(y.min1,y.max1),col="dark grey",lwd=1.5)
#grid
lines(c(25,25),c(y.min1,y.max1),col = "lightgray", lty = "dotted")
lines(c(75,75),c(y.min1,y.max1),col = "lightgray", lty = "dotted")

lines(c(x.min1,x.max1),c(25,25),,col = "lightgray", lty = "dotted")

lines(c(x.min1,x.max1),c(-25,-25),,col = "lightgray", lty = "dotted")

points(compass,xlab="",ylab="",col="dark red",pch=rownames(compass))
points(anchorsplot,xlab="",ylab="",col="dark red",pch=19,cex=0.6)
points(sfish,xlab="",ylab="",col=cols,pch=19)

lp=round(num_fish/2,0)
legend(108,40,rownames(sfish[1:lp,]),col=cols[1:lp],pch=19,cex=0.7,bty="n")

legend(130,40,rownames(sfish[(lp+1):num_fish,]),col=cols[(lp+1):num_fish],pch=19,ce
x=0.7,bty="n")

output=list()
output$scores=sfish[,1:2]
output$ind=sfish[,3]
output$cols=cols
return(output)
}

#---- RADAR PLOTS ---------------------------------------------------

radarplot <- function(dataset) {
# Draw the plot background
cols=rainbow(max(10,nrow(dataset)),start=0, end=.7)
yp = c(-10,-5,5,10,5,-5,-10)
xp = c(0,-5,-5,0,5,5,0)
 plot(xp,yp,type='n',xlim=c(-8,12),ylim=c(-11,11),axes=F,xlab="",ylab="")
polygon(xp,yp,border="dark grey",col="darkseagreen1",lwd=1.5)
polygon(xp*0.7,yp*0.7,border=NA,col="lightyellow")
polygon(xp*0.4,yp*0.4,border=NA,col="mistyrose1")

lines(c(xp[1],xp[4]),c(yp[1],yp[4]),col="dark grey",lwd=1.5)
lines(c(xp[2],xp[5]),c(yp[2],yp[5]),col="dark grey",lwd=1.5)
lines(c(xp[3],xp[6]),c(yp[3],yp[6]),col="dark grey",lwd=1.5)

# Label the background

y.text = c(-11,-5,5,11,5,-5)
x.text = c(0,-7,-7,0,7,7)
text(x.text,y.text,colnames(dataset),col="grey20",offset=0)

legend("right",rownames(dataset),col=cols,lty=1,lwd=2,bty="n",text.col="grey20",cex
=0.8)

# Prepare the data

y.scale = yp[1:6]/100
x.scale = c(0,1,-1,0,1,-1)

fy = y.scale * t(dataset)
fx = x.scale * fy
fx = t(fx)
fy = t(fy)

# plot the data on the background image

for(i in 1:nrow(dataset)) {
polygon(fx[i,],fy[i,],border=cols[i],lwd=2)
}
}