0
I might have say, the sales data from 2017 to 2020, and I wanna use the historical data to build a multiple regression model based on time series to explain the contribution of different independent variables to the dependent one. We all know that when testing model, we have to choose right variable for the final model according to p and t value. So when it comes to p and t, we are talking about hypothesis testing or statistical testing right? So it must involve sample and population cuz p and t value is used to determine if sample statistic can represent the parameters of population right? But I'm really confused about the sample and population in regression model based on time series. I think the historical data from 2017 to 2020 is population(for example, this company who own the data was founded in 2017 and I have all the historical data (from 2017 to 2020). But if it is population, why it has to also involve p and t value when building model. I mean, isn't the t and p is used to measure the probability of sample representing population? But in this case we just have population no sample...