Hi All,
I've recently been trying to use some TDA in R, yet one limitation I have found is in the selection of the "filter values" argument.
My understanding is that they typically use the range of X-axis for the "filter_values" argument, but I've had no success.
Here is a published example:
devtools::install_github("paultpearson/TDAmapper")
devtools::install_github("christophergandrud/networkD3")
library(TDAmapper)
library(igraph)
#DATA
n_x <- c(rep(-.5,101),0.01*(-50:50),rep(.5,101))
n_y <- c(0.02*(-50:50),-0.02*(-50:50),0.02*(-50:50))
d_x <- c(rep(-0.9,101),0.02*(-50:25),0.02*(-50:25),
0.4+sqrt(0.5^2-(0.01*(-50:50))^2))
d_y <- c(0.01*(-50:50),rep(0.5,76),rep(-0.5,76),
0.01*(-50:50))
nd <- data.frame(x=c(n_x,d_x),y=c(n_y,d_y))
plot(nd)
#MAPPER EXAMPLE - FILTER VALUES ARGUMENT
m <- mapper1D(
distance_matrix = dist(nd),
filter_values = c(n_x,d_x),
num_intervals = 10,
percent_overlap = 50,
num_bins_when_clustering = 10)
However, I wondered if someone could help me select the filter_values argument on something like the trial data expressed below, with a range of 1:3 (much more characteristic of my own data).
Would be appreciated.
x1 = rep(1:3, times = 100)
x2 = rep(1:3, times = 100)
x3 = rep(1:3, times = 100)
x4 = rep(1:3, times = 100)
x5 = rep(1:3, times = 100)
DAT <- data.frame(x1, x2,x3,x4,x5)
Many thanks.
Links