# Load necessary libraries
library(funcionalData)
library(BSpline)

# Param1
xVal <- as.vector(dataParam1)
nObs <- dim(dataParam3)[2]

# Create basis expansion system for Param1
fdBasisParam1 <- create.bspline.basis(rangeval = range(xVal), norder=6)
yVal <- as.matrix(dataParam1)
fdParam1 <- Data2fd(argvals=xVal,y=yVal, basisobj=fdBasisParam1, lambda=0)

# Round coefficients to 4 decimal places
round(fdParam1$coefs, 4)

# Plot Param1 data
plot(fdParam1)

# Param2
fdBasisParam2 <- create.bspline.basis(rangeval = range(xVal), norder=6)
yVal <- as.matrix(dataParam2)
fdParam2 <- Data2fd(argvals=xVal,y=yVal, basisobj=fdBasisParam2, lambda=0)

# Round coefficients to 4 decimal places
round(fdParam2$coefs, 4)

# Plot Param2 data
plot(fdParam2)

# Param3
fdBasisParam3 <- create.bspline.basis(rangeval = range(xVal), norder=3)
yVal <- as.matrix(dataParam3)
fdParam3 <- Data2fd(argvals=xVal,y=yVal, basisobj=fdBasisParam3, lambda=0)

# Round coefficients to 4 decimal places
round(fdParam3$coefs, 4)

# Plot Param3 data
plot(fdParam3)

# Param4
fdBasisParam4 <- create.bspline.basis(rangeval = range(xVal), norder=3)
yVal <- as.matrix(dataParam4)
fdParam4 <- Data2fd(argvals=xVal,y=yVal, basisobj=fdBasisParam4, lambda=0)

# Round coefficients to 4 decimal places
round(fdParam4$coefs, 4)

# Plot Param4 data
plot(fdParam4)

# Clustering multivariate functional data with funclust algorithm
allFd <- list(fdParam2,fdParam3,fdParam4)
K <- 3    # Number of clusters
thd <- 0.05  # Threshold for Cantell Scree-Test
hard <- FALSE
nLoop <- 1

cls <- c()
tik <- c()

for (i in 1:nLoop) {
  clustResult <- funclust(allFd,
                          K=K,
                          thd=thd,
                          increaseDimension=FALSE,
                          hard=hard
  )
  tik <- rbind(round((clustResult$tik)*100, 2))
  cls <- rbind(cls,clustResult$cls)
}

# Calculation of class-specific characteristics
uniGroups <- sort(unique(as.vector(cls)))

fbParam2ClsMean <- list()
fbParam3ClsMean <- list()
fbParam4ClsMean <- list()

j <- 1

for (i in uniGroups) {
  obsItems <- which(cls %in% i)
  
  # Mean values
  fdClsMean[[j]] <- mean(fdParam2[obsItems])
  j <- j + 1
}

plot(fdClsMean)

# Plot individual classes
for(i in 1:3){
  plot(fbParam2ClsMean[[i]])
}