Forecasting techniques

There are 3 forecasting techniques I want you to use.

 

  • Proportionate Change Method

-compute the % change in revenue for each year

-compute the mean % change (the average of all the % changes)

-use the mean % change to forecast revenues

-so, if the mean % change = 5% then simply add 5% to the revenues for each year you

are trying to predict. You add 5% to what the actual were.

-you will do this for each year in the time series in order to see how accurate you would

have been throughout the series if you had used the predicted percent change. You can

compare this to the actuals in a graph.

 

  • Three-Year Moving Average

-compute the average revenues for the first 3 years in the series.

-compute the average for the 2nd, 3rd, and 4th years

-keep doing this until you have compute averages for each 3 year period.

-follow the proportionate change method, but use the computed averages in place of the actual.

 

  • Least Square Trend Line

-identify the year in the middle of your data (so if you have data for 2001, 2002, 2003; the central year is 2002). For even numbered years either eliminate year one, or pick between the 2 central years.

-calculate the years from center for each year (so the central year is given a value of 0; the year before the central year is -1; the year before that is -2; etc; the year after the central year is 1; the year after that is 2; etc)

-calculate the slope:

a-multiply each year’s revenue by the years from center values you just calculated

b-sum the results of step a

c-square each years from center value

d-sum the results of step c

e-divide step b by step d

-calculate the level in the central trend:

a-sum the actual revenue for all the years in your dataset

b-divide step a by the total number of years (n) in your dataset

-calculate the trend line:

a-for each year, multiply the years from center value by the slope

b-add the level in the central trend to step a

—this gives you a straight trend line if you plot it on a graph. You will want to make a graph which compares your tend line with your actual.

-forecast future revenues:

-for each outyear you want to forecast you simply take the last value in your trend line and add the value of the slope. So, if you want to forecast 2012 revenues you simply take the 2011 tend line value and add the slope. If you then want to forecast 2013 you take the 2012 value you just calculated and add the slope. Etc.