Synchronise data starting from a fixed threshold

Hello guys,

I'm trying to figure out how to manipulate a matrix to synchronise data of each column based on a given threshold. Below an example of the matrix (12x4). I need to obtain the same matrix but only with values > 1 for a lenght of 5 points (5x4). Generally my data will alwasy have a threshold > 1 and I will always be able to determine a given number of points such as final length will be the same in the end. Columns however may vary, in this case only 4.

I tried some for loop options discussed in another thread but without finding a real solution. I have got stuck.

0.773487342198182  0.600138756438417  0.772170734960123  0.516639735856335
0.864087340442978  0.615542221162260  0.869415217974424  0.537233561755918
0.991682274100864  0.638389426741996  0.982410980592013  0.615436818611677
1.32088921825082  0.706869092358629  1.09369088583133  0.757585637236141
2.28357938862586  0.907713092310730  1.26556377855575  1.00404112351146
4.90532643030186  1.43660651296947  1.90422770765619  1.55522956023284
11.4072812534284  2.77316107773783  4.37097547909669  3.09343285415927
26.1009031962690  6.15011521861756  12.1192439504818  7.51757113064298
56.2552255198974  14.5139540875845  32.2458223148952  19.2312119978824
112.068671806419  33.8800895937913  76.5865546546643  46.6144406361416
204.680987120115  74.3217216672375  160.666680053232  102.442449362564
341.947876642696  149.265403348143  298.912052552757  201.592183165793

[jonas]

I do not understand. Give an example of input and output of a single column.

[Giuseppe]

Input is above in the original message. I would expect this outcome (example first column starting from the first value >1 with length equal 5 points):

1.320889218
2.283579389
4.90532643
11.40728125
26.1009032

Answer 1

[jonas]

With the data given in the comment to this answer, try this script:

A=readtable('data.csv');
A=table2array(A)
A(A<1)=NaN;
A=mat2cell(A,size(A,1),ones(1,size(A,2)))
A=cellfun(@(x)x(~isnan(x)),A,'uniformoutput',false)
A=cellfun(@(x)x(1:5),A,'uniformoutput',false)
out=cell2mat(A)

[Giuseppe]

Thanks. Though, they do increase but at some point they will decrease untill they will reach again a value close to zero. Any idea on how to solve that?

[Giuseppe]

Actual file attached. I think 50 points after the threshold (>1) should give the same length for all columns. This to avoid errors (e.g. mismatch, etc.)

[jonas]

updated the solution

[Giuseppe]

Cheers Jonas.

This mainly work but I spot some issues related to my needs. If I do A<1 = NaN I may end up with some of the data lost because the whole signal is prettymuch a parabole, therefore, I will have <1 at the beginning of the signal but also at the end. I would need to tell Matlab that <1 should be NaN only for a given length, the rest should not be replaced. Not sure if it makes sense but if you look at the data.csv file you will see what the signals look like. I have given the example of new length = 5 but in the specific case of data.csv that would be around 50 points.

[jonas]

Next iteration. I will allow one for-loop this time :)

A=readtable('data.csv');
A=table2array(A);
A=mat2cell(A,size(A,1),ones(1,size(A,2)))
for i=1:length(A)
    A{i}(1:find(A{i}>=1,1,'first'))=NaN;
end
A=cellfun(@(x)x(~isnan(x)),A,'uniformoutput',false)
A=cellfun(@(x)x(1:5),A,'uniformoutput',false)
out=cell2mat(A)

The difference here is that the loop only removes data up until the first '1'. This means that you can now end up with values less than one in your output, granted that they appear shortly after the first one.

[Giuseppe]

Cheers jonas. That improved the code.