nt_demean

PURPOSE

SYNOPSIS

DESCRIPTION

[y,mn]=nt_demean(x,w) - remove weighted mean over cols
 
  w is optional

  if w is a vector with fewer samples than size(x,1), it is interpreted as
  a vector of indices to be set to 1, the others being set to 0.

 NoiseTools

CROSS-REFERENCE INFORMATION

• nt_demean [y,mn]=nt_demean(x,w) - remove weighted mean over cols
• nt_fold y=fold(x,epochsize) - fold 2D to 3D
• nt_unfold y=nt_fold(x) - unfold 3D to 2D
• nt_vecadd y=nt_vecadd(x,v) - add vector to all rows or columns of matrix
• nt_vecmult y=nt_vecmult(x,v) - multiply all rows or columns of matrix by vector

• nt_cca [A,B,R]=nt_cca(x,y,shifts,C,m,thresh,demeanflag) - canonical correlation
• nt_cca_aad [D,E,R]=nt_cca_match_aad(xA,xB,y,ssize) - calculate metrics for match-mismatch task
• nt_cca_crossvalidate [AA,BB,RR,iBest]=nt_cca_crossvalidate(xx,yy,shifts,ncomp,A0,B0) - CCA with cross-validation
• nt_cca_crossvalidate2 [A,B,RR]=nt_cca_crossvalidate2(xx,yy,shifts) - CCA with cross-validation
• nt_cca_crossvalidate_3 [AA,BB,RR,iBest]=nt_cca_crossvalidate(xx,yy,shifts,ncomp,A0,B0) - CCA with cross-validation
• nt_cca_mm [D,E,R]=nt_cca_mm(x,y,ssize,ldaflag,nccs) - calculate metrics for match-mismatch task
• nt_cca_mm3 [D,E,R]=nt_cca_mm(x,y,ssize,ldaflag,nccs) - calculate metrics for match-mismatch task
• nt_cov_lags [C,tw,m]=nt_cov_lags(x,y,shifts,nodemeanflag) - covariance of [x,y] with lags
• nt_demean [y,mn]=nt_demean(x,w) - remove weighted mean over cols
• nt_demean2 y=nt_demean2(x,w) - remove mean of each row and page
• nt_destep [y,stepList]=nt_destep(x,thresh,guard,depth,minstep) - remove step glitch from MEG data
• nt_detrend (original) [y,w,r]=nt_detrend(x,order,w,basis,thresh,niter,wsize) - robustly remove trend
• nt_detrend [y,w,r]=nt_detrend(x,order,w,basis,thresh,niter,wsize) - robustly remove trend
• nt_dprime [d,e]=nt_dprime(x,y,jd_flag) - calculate d' (discriminability) of two distributions
• nt_eyeblink [y,z,mask]=nt_eyeblink(x,eyechans,nremove) - project out eyeblinks
• nt_idx_disp nt_idx_disp(name,field,explainflag) - display contents of index file
• nt_idxx nt_idxx(fname,p) - create an index file to summarize large data file
• nt_inpaint function y=nt_inpaint(x,w) - weighted interpolation based on correlation structure
• nt_iplot [hh,ii]=nt_iplot(fname) - plot data file based on index
• nt_outliers [w,y]=nt_outliers(x,w,thresh,niter) - detect outliers based on weighted correlation structure
• nt_regw (original) [b,z]=nt_regw(y,x,w) - weighted regression
• nt_regw [b,z]=nt_regw(y,x,w) - weighted regression
• nt_sca [M,y,score,proportion]=nt_sca(x,ncomp) - shared component analysis
• nt_sgram [s,f,t]=nt_sgram(x,window,noverlap,nfft,sr,flags) - spectrogram
• nt_sns y=nt_sns(x,nneigbors,skip,w) - sensor noise suppression
• nt_split [idx,score_vector,score]=nt_split(x,depth,thresh,guard,minstep) - split time series into intervals
• nt_split_jd [idx,score_vector,todss]=nt_split_dss(x,thresh,depth) - segmentation based on joint diagonalization
• nt_star [y,w,ww]=nt_star(x,thresh,closest,depth) - sensor noise suppression
• nt_star2 [y,w,ww]=nt_star2(x,thresh,closest,w) - sensor noise suppression
• nt_tsr [y,idx,w]=nt_tsr(x,ref,shifts,wx,wref,keep,thresh) - time-shift regression (TSPCA)
• nt_zapline [y,yy]=nt_zapline(x,fline,nremove,p,plotflag) - remove power line artifact

SOURCE CODE

0001 function [x,mn]=nt_demean(x,w)
0002 %[y,mn]=nt_demean(x,w) - remove weighted mean over cols
0003 %
0004 %  w is optional
0005 %
0006 %  if w is a vector with fewer samples than size(x,1), it is interpreted as
0007 %  a vector of indices to be set to 1, the others being set to 0.
0008 %
0009 % NoiseTools
0010 
0011 if nargin<2; w=[]; end
0012 if nargin<1; error('!');end
0013 
0014 if iscell(x)
0015     if ~isempty(w); error('!'); end; % not implemented
0016     for iCell=1:numel(x)
0017         [x{iCell},mn{iCell}]=nt_demean(x{iCell}); % ! behaves like nt_demean2
0018     end
0019     return;
0020 end
0021 
0022 if ~isempty(w) && numel(w)<size(x,1)
0023     w=w(:);
0024     % interpret w as array of indices to set to 1
0025     if min(w)<1 || max(w)>size(x,1); 
0026         error('w interpreted as indices but values are out of range');
0027     end
0028     ww=zeros(size(x,1),1);
0029     ww(w)=1;
0030     w=ww;
0031 end
0032 
0033 
0034 if size(w,3)~=size(x,3);
0035     if size(w,3)==1 && size(x,3)~=1;
0036         w=repmat(w,[1,1,size(x,3)]);
0037     else
0038         error('W should have same npages as X, or else 1');
0039     end
0040 end
0041 
0042 [m,n,o]=size(x);
0043 if ndims(x)==3; x=nt_unfold(x); end
0044 
0045 if isempty(w);
0046     
0047     mn=mean(double(x),1);
0048     x=nt_vecadd(x,-mn);
0049     
0050 else
0051     
0052     w=nt_unfold(w);
0053     
0054     if size(w,1)~=size(x,1)
0055         error('X and W should have same nrows'); 
0056     end
0057     
0058     
0059     if size(w,2)==1;
0060         mn=sum(nt_vecmult(double(x),w),1) ./ (sum(w,1)+eps);
0061     elseif size(w,2)==n;
0062         mn=sum(x.*w) ./ (sum(w,1)+eps);
0063     else
0064         error('W should have same ncols as X, or else 1');
0065     end
0066 
0067     %y=bsxfun(@minus,x,mn);
0068     x=nt_vecadd(x,-mn);
0069     
0070 end
0071 
0072 x=nt_fold(x,m);
0073