Fitting every fifth datapoint one after another

Hello!

I want to do a linear regression for a bunch of data, using every fifth or third (maybe tenth) Datapoint. The data will be vary in size and therefor I need to do a function that varies as well.

I did the poly1-fit to the first 13 data already by typing it for each fit. Can you think of any way to solve it via function maybe a loop to solve this seccessively for any size of x- or y-datapoints.

I will be happy for any kind of response that helps me to solve this problem.

Thanks a lot!

The result should look like this for more datapoints:

Answer 1

[dbmn]

I have a solution that avoids the loops altogehter, but might not be as readable

n = 5; % Number of datapoints
x1 = reshape(x, n, []); % please make sure, that your number will be divisable into n Elements
y1 = reshape(y, n, []);
a = (x1-x1(1,:))\(y1-y1(1,:)); % we use mldivide here and correct for the origin
a = a(1,:);
b = y1(1,:) - a.*x1(1,:); % here we shifted back

b is the offset and a the slope of your curves

[dbmn]

oh and for the plotting part you can use

figure()
hold on
hdl1 = plot(x1, y1, 'b.');
hdl2 = plot(x1, a.*x1+b, 'r-');

to avoid loops

[J.P.]

Hi! Thanks for your help. But your solution gives this Error:

Error using - Matrix dimensions must agree.

Error in polyfit_Inet (line 6) a = (x1-x1(1,:))\(y1-y1(1,:)); % we use mldivide here and correct for the origin

[J.P.]

Okay, I changed the parts a little bit and it works!!!

n = 6; % Number of datapoints
x1 = reshape(x, n, []); % please make sure, that your number will be divisable into n Elements
y1 = reshape(y, n, []);
a = (x1(1)-x1(1,n))\(y1(1)-y1(1,n)); % we use mldivide here and correct for the origin
a = a(1,:);
b = y1(1,:) - a.*x1(1,:); % here we shifted back
figure()
hold on
hdl1 = plot(x1, y1, 'b.');
hdl2 = plot(x1, a.*x1(2,:)+b, 'r-');

But what I get now is a fit including 6 regression lines, which are shifted in parallel.