# Autoregression Model

I am new to statistics and math in general as I started at such a late age. I am trying to come up with a model that can predict future numbers using past numbers based on multiple time series that fit into one another. Each time series should operate somewhat independently of the other but also should be dependent on one another. How can I set up a model that will encompass this information that has boundaries. For instance. If I used a time series of 1 hour with the number of data points being 20 with boundaries of 2 standard Deviations above and below the moving average. Now use a time series of a half hour that also uses 20 data points with boundaries of 2 std dev.above and below, how could I build a model that will run that will try to bring these systems into equilibrium? Keeping in mind that the 20 data points in the hourly series is equal to twenty hours but the 20 data points in the half hour series is only 10 hours. Any and all help would be appreciated. Thanks.

• Can you provide more information, or a simple example? This is rather abstract & hard to follow. – gung - Reinstate Monica Jun 24 '15 at 0:33
• Im sorry for the confusion. i will try to explain it differently. Lets say i have a moving average of 20 numbers. I want to find the next numbers in the series. Lets also assume that these twenty numbers came in each hour for 20 hours. Lets say they have a range that is 2 standard deviations above and below the 20 period moving average. The boundaries would change with each new number. The new number would be trying to get to equilibrium, that is become the moving average.So at some times, it is above the moving average and sometimes it is below the moving average. – clint Jun 24 '15 at 1:11
• The closer you get to the moving average above and below the closer the new numbers start becoming the moving average. So for example if we used 10 numbers for a series and the were in a specific order such as 2,3,3,4,4,5,6,7,8,9, the moving average would be around 5. However if the next number was 8, the new moving average would be closer to 6 since you would be getting rid of the 2 at the beginning. The numbers would start moving down towards the moving average when it is above it and up towards it when it is below it. – clint Jun 24 '15 at 1:11
• The boundaries above and below would make the number pivot back the other way. So the question is, when I add other time series into that, such as a half hour series and a 15 minute series, etc. how can I get those series to work in relation to the hour? Each new number should be based on coming to equilibrium. However, equilibrium in one series might not be equilibrium in another. Lets say we are in equilibrium in a 15 and 30 min series but below the moving average on the larger one such as a hour, it would push the smaller ones out of balance upward and also push the upper limits up. – clint Jun 24 '15 at 1:13
• So the larger one gets closer to equilibrium but the smaller ones are now out of balance and need to reverse when the number hits the upper limit of the larger hr time series. I hope this makes a little more sense. If not, I will try again. Thanks – clint Jun 24 '15 at 1:13