repflag=true;

hi there.. i am converting matlab codes which have "repflag=true" i don't understand this kindly someone guide me...

thanks in advance.

[David Goodmanson]

Hi Sher,

impossible to say since you have not provided any context. Regardless, what you could do at least as well as anyone on this site is scan down through the rest of the code until you find

if repflag      % or possibly   if repflag==true
  blah
  blah
  blah
end 

and try to scope out what it does.

[Sher Akbar]

okay i can understand ... so this is the full coding what i am asking for...

#########################

#####Recursive filter on prices

function [ dataf, spikes, drops ] = of_rfp(data, param)
    if(nargin < 2 || isempty(param))
        ns = 3;
    else
        ns = param;
    end

    outrep = mean(data);

    spikes = zeros(size(data));
    drops = zeros(size(data));
    dataf = data;
    repflag = true;
    while(repflag)
        spikestmp = logical(dataf >= mean(dataf)+ns*std(dataf));
        dropstmp = logical(dataf <= mean(dataf)-ns*std(dataf));

        if(nnz(spikestmp))
            dataf(spikestmp) = outrep;
        end
        if(nnz(dropstmp))
            dataf(dropstmp) = outrep;
        end

        if(~nnz(spikestmp) && ~nnz(dropstmp))
            repflag = false;
        end

        spikes = spikes + spikestmp;
        drops = drops+ dropstmp;
    end
    spikes = (spikes == 1);
    drops = (drops == 1);
end

please give me any suggestion .. thanks for replying.....

[David Goodmanson]

is 'data' a vector, or is it a matrix?

[Walter Roberson]

The code appears to be looping replacing outiers with the mean of the data, until finally there are no more outliers.

[Sher Akbar]

it is a vector.. but the problem is i want to replicate this command in R...

So, what changes i should make??

[David Goodmanson]

HI Sher,

best to figure out what it's doing, then write a new version in R. The comment says it's recursion, but it's really iteration.

Using ns = 3 as an example, then as Walter mentioned, take the outlier measurements that lie more than 3 standard deviations away from the mean, and replace them by the mean. That makes a new data set. Then find a new mean and standard deviation and repeat the process until there are no more outliers. Why someone would use that process is a question worth asking, if you don't know already.

During the process, count up the outliers on each side of the mean. The spikes = (spikes ==1) statement seems to be in order to avoid double counting any outliers, same for drops.

[Sher Akbar]

thank you for your reply...

does it make sense to use repeat and while loop both at a time??

should i use it?

[Sher Akbar]

can you please give me an idea about any looping algorithm??

which loop has to be used now?

while or something else..???

[David Goodmanson]

Hi Sher, I don't know R and redoing this in R is not really within the scope of this website.

[Sher Akbar]

okay thank you so much for your quick response.. :)