When analyzing time series (data indexed by time), there are several key concerns that exist when modelling the data. What are some of those concerns? How are they addressed using more classic statistical / econometric methods of inference & learning?