Trend/Forecast Sales Data

[stirish]

Good Evening!

I am in the process of trending or forecasting sales data for a project at work. I have been asked to determine, on our current trajectory where we will end the year in 4 different sales categories. The information I have to make my determinations is monthly totals for each of the categories for the last 6 years. I need to trend or forecast out to year end in December 2017. Since I did this "wrong" the last time and just annualized the data, I figured I would reach out to the experts on this forum (as I was recommended by a friend!)

Like most I am behind the 8 ball a bit with this so any help is GREATLY appreciated.

Thank you!

Tom

 

[Rob65]

Hi Tom,

you seem to have a lot of data (6 years), so one thing you could think about doing is using a centred 12-month moving average. For example, your first point would be the average of the months 1-12, the second the average of 2-13 etc. This will give you a rolling annual average hat is updated every month. All you really need to do then is to fit a regression line through the point that you have determined and project it to the point in time of relevance to your project.

The logic of using a 12-moth moving average is twofold. First, each data point encapsulates a full year- therefore it removes any seasonal variation. Second, individual spikes are damped and shouldn't have a significant skewing effect that you might get with the raw data points.

To be thorough, you should probably determine if the trend observed is statistically significant, but unless you are competent in inference statistics, then an eyeball will probably be sufficient.

Hope this helps.

Regards,

Rob