# Autocorrelation with additional binary predictor

I have the data in the following format:

      release1 release2 release3 release4
word1     500      460     370      0
word2     0        10      30      100
word3    350       350    500      400


where release t is simply a time t. For each t I have several binary variables, such as IsRefactoring, IsBugFix, IsFeatureAddition. I know that autoregression utilize lagged variables t-1 to predict t, but I've never seen someone used other independent variables in addition to that. I'd like to know how to include those binary variables as predictor variables into my analysis and if there's a statistical method for that.