How to do OlS to get the residual?

The model is y=a+bt (t was the time,I changed it into id,y means the GDP per captital of US,a means the intercept) Here it is better that y could be transferred into log formal I am aimed at getting the minimum of the (y-yhat)^2. Further more ,I need to write code about it. But I have no idea where and how I can start it Could you tell me the code and logicality. Many thanks.

Answer 1

[Star Strider]

You may want to look at How can I do log-linear de-trending in data?

[Wong Yi]

How to build linear function as I said ？

[Wong Yi]

Do you know what's meaning of code

B0 = [1; -1; 1]*0.001

What is 1,-1,1,0.001?

[Star Strider]

A linear regression would be:

B = [ones(size(d(:,1))) d(:,1)]\d(:,2);

The fitted regression is then:

reg = [ones(size(d(:,1))) d(:,1)]*B;

Do that for both nations, as I did in the loop. This requires that you change ‘d(:,2)’ to ‘d(:,3)’ in the second regression. As I understand what you are doing (no promises), everything else is unchanged.

[Wong Yi]

What 's the meaning of d

[Star Strider]

In How can I do log-linear de-trending in data? my original Answer addressing this problem, ‘d’ is the data matrix returned by the xlsread call, where ‘d(:,1)’ was the time vector, ‘d(:,2)’ USA, and ‘d(:,3)’ JPN.

Please see that Answer for a comprehensive discussion of the problem and my approach to the solution.