Thanks,Following your explanation i have through bi conductor but still unable
to figure out how to group. for the data, my data takes values 0,1,2 hence
simulating from a multinomial distribution such as (rmultinom(10, size = 3,
prob = c(0.1,0.2,0.8))) gives 10 variables from three individuals assume this
variables are ordered.I am to develop an r algorithm that will develop groups
as follows;
i) Calculate the correlation between successive variables and correlation
matrix for all the variables
ii) if r(i) is the maximum correlation coefficient in the successive variable
correlation then my first group has variables [x(i-k ), ...x(i),... x(i+k )].
this gives me group one of size 2k+1. note that variables are picked on that
line and order is maintained.iii) the max r is removed from the list of
successive correlation between variables, but we don't recalculate the
correlation since this will destroy the array.
iv)check for remaining correlation btn successive variables relative to the
bigger correlation matrix and select other groups. these groups don't need to
be of same size and they must not overlap.
e.g,if for instance x1, x2,......x10 are my 10 variables. i calculate
correlation btn successive variable. Say the max correlation is Cor(x4,x5) i
take group one to have values (x2,x3,x4,x5), my data remains,
x1,-,-,-,-,x6,x7,x8,x9,x10. Clearly x1 will form a group and the algorithm has
to check for other groups form x6,x7,x8,x9,x10 based on max r and also the
correlation matrix.
Note that i have large number of variables. i don't need the groups to overlap
i will have divided in to disjoint groups but maintaining the order.
kindly assist me see how i can get this done, thanks
On Thursday, November 13, 2014 1:50 PM, Cox Lwaka <clw...@yahoo.com> wrote:
I have a bit of trouble here to program in r. I am anew user but i really
enjoy working with it.I have a large number of variables in a matrix that are
arranged sequentially on a line (chromosome). This order has to be maintained
whatsoever. I am to develop an r algorithm that will develop groups as
follows;i) Calculate the correlation between successive variables and
correlation matrix for all the variables
ii) if r(i) is the maximum correlation coefficient in the successive variable
correlation then my first group has variables [x(i-k ), ...x(i),... x(i+k )].
this gives me group one of size 2k+1. note that variables are picked on that
line and order is maintained.iii) we check for remaining correlation btn
successive variables relative to the bigger correlation matrix and select
other groups. these groups don't need to be of same size and they must not
overlap.i will have divided in to disjoint groups but maintaining the
order.kindly assist me see how i can get this done, thanks
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
