Add t-SNE method of reduction clustering if usable

[trinker]

https://lvdmaaten.github.io/tsne/
https://datascienceplus.com/multi-dimensional-reduction-and-visualisation-with-t-sne/

[trinker]

pacman::p_load(Rtsne) 

iris_unique <- unique(iris) 

tsne_out <- Rtsne(as.matrix(iris_unique[,1:4])) 
plot(tsne_out$Y,col=iris_unique$Species)

[trinker]

In R, the TSNE algorithm is available in two packages: TSNE and RTSNE. TSNE is a R implementation of the algorithm and can take an incredibly long time to map large data sets. RTSNE is a wrapper around the C++ implementation of the algorithm and can handle larger data sets with low computation time. The graph below compares the various implementations and shows that RTSNE is 35 times faster than the TSNE implementation.

http://simranmetric.com/practical-considerations-for-using-tsne/

[trinker]

Ncols or rows may limit..see:
https://stackoverflow.com/questions/45022212/why-is-this-r-code-for-a-t-sne-analysis-running-so-slowly