I must to predict the values of an outcome variables (economic return) for a number of units, where each unit represents a specifical shopping point . This variable is a fuction of a number of categorial variables, of a number of sales variables and of a lag fuction of the outcome itself.
Now I have a time variable as well, which identifies N different time levels. In other words I have an N replication of the information above for all of the shopping points, except for the Nth level for which I don't have any value for the outcome variable.
Do you have an idea on which technique I should use to predict the values for the N-th level of the outcome variables based on the observations available for the previous time levels?
Thank you in advance to everybody!