TestPackage.Rmd

#Install R3.4.0
#Install Rtools34.exe > https://cran.r-project.org/bin/windows/Rtools/history.html
#Use R 3.4 > change in Tool s > Global Options > General >R Version
```{r}
#install.packages("rmarkdown")
#update.packages(ask = FALSE)
#install.packages("Rcpp")\

install.packages("reticulate")


```
#Make sure built under R 4.0.0
#Restart Rstudio


---
title: "testing modified density clustering code"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```



----------------------------------------------------------------------------------

```{r}

#writeLines(strsplit(Sys.getenv("PATH"), ";")[[1L]])
library(Rcpp)
library(reticulate)
source("RcppExports.R")
source("plotDensityClust.R")
source("densityClust.R")
#sourceCpp("RcppExports.cpp")
sourceCpp("localDensity.cpp")
sourceCpp("findDistValueByRowColInd.cpp")
sourceCpp("distanceToPeak.cpp")
```
```{r}

```

---------------BCI Data Prep--------------------------

#Read in the BCI Data
```{r}
Colnames = c('X','Y','Weight')
BCI_3_Null <- read.csv(file.choose(), header = FALSE)
#dimnames(BCI_3_Null) <- NULL
```

```{r}
BCI_Weights <- as.vector(BCI_3_Null[,3])
BCI_Noweights <- as.vector(BCI_3_Null[,1:2])
BCIDist <- dist(BCI_3_Null[,1:2])
```


---------------BCI Data with Rodriguez Clustering--------------------------

#cluster with non-gaussian distance

```{r}
BCIClust <- densityClust(BCI_Noweights, BCI_Weights, BCIDist, gaussian=TRUE, verbose = TRUE)
```


#get validation metrics fro rho/gamma combinations
```{r}


BCIChart1 <- findCluster_validationChart(BCIClust, DBCV = FALSE, status = FALSE)
View(BCIChart1)
```

#Plot the BCI Data with Chosen Clusters
```{r}


plot(BCIClust) # Inspect clustering attributes to define thresholds
BCIClust <- findClusters(BCIClust, rho=4.01, delta=0.21, verbose = FALSE)
#BCIClust <- findClusters_DBCV(BCIClust, rho=338, delta=0.21, verbose = FALSE)
BCIRAW = plotRAW(BCIClust)
BCIMDS = plotMDS(BCIClust)
BCITSNE = plotTSNE(BCIClust)

#split(iris[,5], BCIClust$clusters)
```











---------------BCI Weights as third dimension--------------------------

```{r}
BCI_Weights <- rep(1,224)
BCI_Noweights <- as.vector(BCI_3_Null[,1:3])
BCIDist <- dist(BCI_3_Null[,1:2])

```


```{r}
BCIClust <- densityClust(BCI_Noweights, BCI_Weights, BCIDist, gaussian=TRUE, verbose = TRUE)
```



```{r}
BCIChart2 <- findCluster_validationChart(BCIClust, DBCV = FALSE, status = FALSE)
View(BCIChart2)
```

```{r}
plot(BCIClust)
BCIClust <- findClusters(BCIClust, rho=2.31, delta=0.101, verbose = FALSE)
BCIRAW = plotRAW(BCIClust)
BCIMDS = plotMDS(BCIClust)
BCITSNE = plotTSNE(BCIClust)
```










---------Test Iris Code/ Old Code---------------



```{r}
iris <- iris
noweights <- rep(1,150)
irisDist <- dist(iris[,1:4])
irisClust <- densityClust(iris[,1:4], noweights, irisDist, gaussian=TRUE)
plot(irisClust) # Inspect clustering attributes to define thresholds




```

```{r}
irisChart <- findCluster_validationChart(irisClust)
```

```{r}
irisClust <- findClusters(irisClust, rho=0.01, delta=0.74)
raw = plot(irisClust)
mds = plotMDS(irisClust)
tsne = plotTSNE(irisClust)


#split(iris[,5], irisClust$clusters)
```


Create A Formatted Dataset where each value is multiplied to fit it's num of observations
```{r}

total = nrow(BCI_3_Null)

BCI_3_Format <- data.frame("Cord1" =  double(), "Cord2" = double())

for (i in 1:total){
  for (j in 1:(BCI_3_Null[i,3])){
    newRow <- BCI_3_Null[i,1:2]
    BCI_3_Format<- rbind(BCI_3_Format, newRow)
  }
}
    
```



```{r}
same = 0
frame = BCI_3_Null
total = nrow(frame)
for (i in 1:total){
  for (j in 1:total){
    if ((i != j) & (frame[i,1] == frame[j,1]) & (frame[i,2] == frame[j,2])){
      same = 1
      cat("same value detected at rows ")
      i
      j
    }
  }
}

if (same == 0){
  cat("no rows same ")
}
```

#make sure distance has no 0 values
```{r}
BCIDist <- dist(BCI_3_Null[,1:2])


d = as.matrix(BCIDist)


same = 0
total = nrow(d)
for (i in 1:total){
  for (j in 1:total){
    if ((i != j) & (d[i,j] == 0.0)){
      same = 1
      cat("same dist detected at rows ")
      i
      j
    }
  }
}

if (same == 0){
  cat("no dist = 0 ")
}
```



#cluster without gaussian distance
```{r}
BCIDist <- dist(BCI_3_Null[,1:2])



BCIClust <- densityClust(BCIDist, gaussian=FALSE)
plot(BCIClust) # Inspect clustering attributes to define thresholds

BCIClust2 <- findClusters(BCIClust, rho=5, delta=0.15, verbose = TRUE)
plotMDS(BCIClust2)
#split(iris[,5], BCIClust$clusters)

```