I have a question for LASSO algorithm. From my understanding, LASSO algorithm can help me to determine among a lot of variables, which of them are useful and which of them are not, but existing LASSO regression method assumes independencies within variable. I want to apply this method on my analysis for time series and here's the task I need to perform:
I have 70 exogenous variables that are all continuous time series (for example, max / min air temperature, soil temperature, etc.), my target variable is groundwater head, which is also time series on a monthly basis, and I'd like to find out which one/ones of my exogenous variables can be predictive to my target variable. Within each exogenous variable, there might be a lagged effect, which can be up to 6 months, so in total I have 420 time series to predict my target variable. My response variable can also be predictive to itself, and can apply an AR(p) model on itself. I wonder if there is an R package that can do this kind of analysis?
Thanks a lot for your suggestions!
