San Jose State University Seasonalized Forecasts Discussion

Seasonalized Forecasts

For the seasonalized forecasts, I had to first plot down all the sales from the 2016-2020 table in order to be able to calculate the moving average. The moving average of my table was calculated by using the I column of the beginning table times the q3 index. We use the moving average because we want to smooth out our price data and create a constantly updated average price. Then we calculate the 5,3,1 column by multiplying the K column by the same q3 index as the last column. 5,3,1 is calculated by picking the numbers we want (5th, 3th and 1st), weighting them appropriately to their given number, multiplying each number by their weight and adding up the resulted values to get the average. We calculate the “a = .7” column by multiplying the K column by the q1 index. Lastly, we calculate the “a = .2” column by multiplying the L column from the above table by the q1 index.

Best Forecast by MAD

The Least Squares method seems to be the best forecast method because it gives us data that is the most similar to our original sales. All other methods work decently well but the Least Squares made it the easiest to see the future sales and patterns. We determine the best forecast by observing which method gives us the most accurate results. We must use our common sense to determine which result is best at the very end. However, the determination of the best method depends on the availability of historical data, amount of accuracy desired and the time period of the forecast. In our case, the Least Squares method gave us the best and most sensible data so we think it was the best method for our project.