I have been using basic smoothing techniques such as moving average, median smoothing and exponential smoothing to generate forcasts for the time series in the attached excel file. I would like to know if more advanced smoothing techniques would give better results. The columns in the file are a time series of performance figures for 16 football teams over about 4 years. The best results I've achieved so far, are with exponential smoothing with w=0.2 . If more advanced techniques would improve forcasts, please give a brief explanation of the type of technique used.
There are several points I would like to clarify for this problem:
First of all, we need to understand the difference between smoothing techniques and forecasting techniques. Normally speaking, the smoothing techniques are used to filter out the noises of the data to get the underlying pattern of the data, whereas forecasting techniques are used to make forecast from the given data set. One has to be really careful when using smoothing techniques to make a forecast. I will discuss this in more detail in the next section. Here I give an example of using forecasting techniques to model this time series data:
The basic time series forecasting model is the ARIMA model. Here I take the ...
The smoothing techniques for moving averages are determined.