regression learner equation output

I've generated an svm prediction module using Matlab's regression learner. I now wan't to use that module to predict new data. My system is a power or electricity load forecast that takes the previous load data in addition to previous weather data to predict the load on the next hour. I know that the data should include predictors (features) and response. the predictors are the training data and the response is the expected output. I wan't my system to take say the last 10 samples of previous data and use it to predict the next hour. How can I establish that in Matlab? I mean, how does it know that this portion (10 samples) of the predictor vector is supposed to give response #x in the response vector? When I fed the regression learner a big vector of inut data and it's corresponding expected response I got the following response with RMSE of 1.18 which is good. I just want to understand how its controlled for further use.

I guess to wrap what I mean by all of that is how to represent my system in a function that will be called to predict the next hour based on the load from the past few hours' data.

thanks

[NM]

Hello Isra, I am doing work similar to yours. I have 8 parameters as predictor inputs for load forecasting using SVM. But when I train my data, my RMSE value is around 300. I don't understand why. Do you know why?

[Isra Haroun]

how did you setup your input output matrix for the regression learner? mine contained wheather data (1 column) historical hourly forecast(1 colum), expected hourly forecast (which is shifted version of column 1)

[NM]

My input Parameters are temperature, week day, date, type of day, previous hourly forecast, previous week forecast. What Advanced SVM Options did you use for training? Box Constraint Mode value, Kernel scale mode value etc?