1D interpolation by y = ax+b

I have x and y data. I need interpolate (fit) data by y = a*x + b

Answer 1

[Star Strider]

To set the start point at the origin (0,0), calculate the slope as:

slope = x(:) \ y(:);

for example:

x = 0:0.1:10;
y = rand(size(x));

slope = x(:) \ y(:);

figure
plot(x, y, '.b')
hold on
plot(x, slope*x, '--r')
hold off
grid
xlabel('x')
ylabel('y')

.

[Jaroslav Zadnik]

thank you, but I want to also know parametr a from y = ax+b

[Star Strider]

My pleasure!

That is the ‘slope’ variable in my code example.

Another way of writing that would be:

a = x(:) \ y(:);

producing the same result, this time assigning it to ‘a’.

[Jaroslav Zadnik]

Thank you!

I have some other troubles:

1) For lower curves is slope NaN

2) Where the equation y = ax+b is hidden ?

[Star Strider]

My pleasure!

1) The slope would be NaN only if both the ‘x’ and ‘y’ variables are vectors of zeros. Otherwise, the slope would be between 0 and ±Inf.

2) I do not understand. If you want to display the linear equation, use the text function. (I can help with that, however I need more information as to what you want to do.)

.

[Jaroslav Zadnik]

I would like to write it like this:

F1 = fit(x(1,:)',y','0.5+0.5*cos(2*pi*x/96-delta)','lower',[0],'upper',[2*pi],'startpoint',[pi]);

where I can write the equation directly

[Star Strider]

I do not have the Curve Fitting Toolbox (since I do not need it for what I do in my research). This requirement was not implied or stated in the original Question you posted.

Without your data, I cannot code a similar function using the Statistics and Machine Learning Toolbox or the Optimization Toolbox functions, both of which I have and frequently use. I would need more information to provide an appropriate response.