How to use interp1 in my situation

Hi

I've a .dat file. I'm supposed to make a function that reads the file, extracts 2/3 parameters and interpolate them. So far I've done this:

function [ output1 , output2 ] = functionname( filename , time )
%Define my delimiter
delimiter = '\t';
% %s are columns included, %*s are columns excluded.
formatSpec = '%s%s%*s%*s%*s%*s%*s%*s%*s%*s%*s%s%[^\n\r]';
%Open the file
fid = fopen(filename,'r');
%Read columns from file using textscan.
dataArray = textscan(fid,formatSpec,'Delimiter',delimiter,'HeaderLines',1,'ReturnOnError',false);
%close the file
fclose(fid);
%Convert data from strings to numbers.
res = [datenum(dataArray{1},'yyyy-MM-dd HH:mm:ss'),str2double(dataArray{2}),str2double(dataArray{3})];

%MOOORE

end

Want I've done is: Opened file, saved 3 parameters in matrix called res, and closed file. Now I need to interpolate. The first column in res is original time of measurements. Measurements are taken every 10 minutes. My input 'time' is every hour. So I've to interpolate from the original measurements time to my 'time'. (In my assignment it's recommended to use interp1)

Measurements were taken over a year, so the matrix res has 50000+ lines. Sometimes measurements weren't taken, and are insted called NaN, I need to do something so this doesn't 'cause problems with other functions.

I hope someone can help me. If you need additional info I'll give it to you ASAP! (I use Matlab 2013b)

[dpb]

Since it's an assignment, I'll not provide a direct solution but offer a couple of hints.

To avoid the NaN issue unless there's another directive, since the data is collected so frequently and unless there are large gaps when the measurement device was unavailable, I'd just remove those entries from the vectors to use in interp1. That way you'll eliminate them from showing up; will just interpolate between the remaining data. isnan will be useful here.

Then, the rest if pretty-much straight forward -- convert your time vector to datenum as well and use it as the XI argument and Voila! magic happens! :)

[hmtrondheim Tróndhiem]

Can you tell me what goes wrong here:

mask = ~isnan(res(:,2));
res(:,2) = interp1(res(mask,1), res(mask,2), time,'nearest');

I get "The grid vectors are not strictly monotonic increasing." Both res(:,1) and time go from 2013-01-01 to 2013-31-12 just with different pace. Shouldn't they be monotonic increasing?

[dpb]

Seems like should be from that description, yes...

I see you took the mask from the 2nd column of the data values only. Not seeing your data, I wonder from your initial description if you might not need to use the logical 'or' of both column 2 and 3?

Also check the returned datenum values...try

datestr(res(1))

and make sure you get a sensible date string back.

Can check for certain on the order if you try

all(diff(res(mask,1))>0)

(the difference between all successive datenums is >0) If that indeed comes back as 0 (false), then there is definitely something going on in the date conversion.