Data Interpolation using interp1

Hi

I am trying to interpolate different data sets, so that they all have the same size for my further analysis.

        maxLength = max([length(unique_expperimental_dis), length(unique_experimental_force), length(abaqus_force)]);
        xFit = 1:maxLength;
        experimental_dis_interp = interp1(1:length(unique_expperimental_dis), unique_expperimental_dis, xFit);
        size('experimental_dis_interp')
        experimental_force_interp = interp1(1:length(unique_experimental_force), unique_experimental_force, xFit);
        size('experimental_force_interp')
        abaqus_force_interp = interp1(1:length(abaqus_force), abaqus_force, xFit);
        size('abaqus_force_interp')

But wenn I read out the sizes of the different data sets, they do not match eachother. Any suggestions what I am doing wrong?

Barbara

[Star Strider]

Reference: Error using interp1

Answer 1

[Walter Roberson]

        maxLength = max([length(unique_expperimental_dis), length(unique_experimental_force), length(abaqus_force)]);
        xFit = 1:maxLength;
        experimental_dis_interp = interp1(1:length(unique_expperimental_dis), unique_expperimental_dis, xFit);
        size(experimental_dis_interp)
        experimental_force_interp = interp1(1:length(unique_experimental_force), unique_experimental_force, xFit);
        size(experimental_force_interp)
        abaqus_force_interp = interp1(1:length(abaqus_force), abaqus_force, xFit);
        size(abaqus_force_interp)

You were calculating the size() of the character vectors, not the variables.

Caution: you have not turned on extrapolation, so the above code is equivalent to padding the short vectors with NaN.

[Barbara Schläpfer]

Thank you for your quick answer. Do I understand that right, that I will have to change my xFit ?

[Walter Roberson]

If padding with NaN was not your intention, then you need to be more clear about your desired outcome.

If you have two variables in which there is a linear relationship between the two, so (say) 2/3 of the way through the range of one should correspond to 2/3 of the way through the range of the other, then you can scale between them:

fraction_through_first_range = (value_in_first_range - minimum_of_first_range) / (maximum_of_first_range - minimum_of_first_range);
projected_into_second_range = minimum_of_second_range + (maximum_of_second_range - mininum_of_second_range) * fraction_through_first_range

Is it really the number of different entries in a group that matters to you, or are you wanting to project by portion of the way through the respective range?

linspace(unique_expperimental_dis(1), unique_expperimental_dis(end), maxLength)