09-spectral-clust.Rmd

---
title: "Spectral Clustering"
author: "Kundan K. Rao"
date: "01/12/2021"
output: html_document
---

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

Spectral clustering
```{r}
# gaussian data
# transposed data frame
data.scaled.trans <- as.data.frame(t(data.scaled))
```

```{r}
library(Spectrum)
spec.clust <- Spectrum(data.scaled.trans,showpca=T,fontsize=8,dotsize=2)
#spec.clust
```

```{r}
# non gaussian data
spec.clust2 <- Spectrum(data.scaled.trans,showpca=T,method=2,tunekernel=T,fontsize=8,dotsize=2)
```
# Density based clustering
## Determining optimum eps value
```{r}
library(dbscan)
dbscan::kNNdistplot(data.scaled, k = 2)
abline(h = 2.47, lty = 2)
```

```{r}
# Compute DBSCAN using fpc package
library("fpc")
set.seed(123)
db <- dbscan::dbscan(data.scaled, eps = 2.47, MinPts = 2)

# Plot DBSCAN results
library("factoextra")
fviz_cluster(db, data = data.scaled, stand = FALSE,
ellipse = FALSE, show.clust.cent = FALSE,
geom = "point",palette = "jco", ggtheme = theme_classic())
```