Given the following data sources and their availability:
-=-=-=- Quarterly -=-=-=- Number of High MIP Users Number of Med MIP Users Number of Low MIP Users
Number of MIPS used by all HIGH MIP USERS Number of MIPS used by all MED MIP USERS Number of MIPS used by all LOW MIP USERS (These two together give me the average MIPS used per user, per class)
-=-=-=- Monthly -=-=-=- Number of Jobs Ran SHORT Number of Jobs Ran MED Number of Jobs Ran LONG (user submitted and regularly scheduled jobs)
-=-=-=- Weekly -=-=-=- Number of MIPS Used in Production Number of MIPS Used in Test1 Number of MIPS Used in Test2 Number of MIPS Used in Test3 (What I am trying to forecast sans Production)
-=-=-=- Annually -=-=-=- Test Schedule Dates
What is the best strategy to do a time series analysis when trying to forecast the weekly MIPS to use? The only thing I can modifiy in a planning effort is the Test Schedule and all the data is only for the Test systems (so I am effectivly blind to the Production metrics outside of it's MIP usage.)
I am guessing a regression analysis of sorts, comments and suggestions?