A few years ago I did some fore­cast­ing work for a com­mon­wealth gov­ern­ment depart­ment and found that they were fore­cast­ing a $5 bil­lion bud­get using the FORECAST com­mand in Excel. Worse, they were fit­ting a regres­sion through only three obser­va­tions and they were not even the most recent observations.

It seems a sim­i­lar thing has hap­pened again. The Vic­to­rian gov­ern­ment is pro­ject­ing water con­sump­tion based on a regres­sion through three obser­va­tions. Ken­neth David­son in today’s Age reports that “The Gov­ern­ment pre­dicts a water sup­ply cri­sis based on run­ning a regres­sion curve through the three drought years of 2004–2006, which shows the reser­voirs that sup­ply Mel­bourne dry­ing up by 2010.” Who pro­duces this non­sense and why doesn’t the gov­ern­ment seek proper advice from pro­fes­sional statisticians?

At least the com­mon­wealth gov­ern­ment did per­ceive that there was a prob­lem, which is how I got involved. But the Vic­to­rian gov­ern­ment seems to have another agenda, and respon­si­ble sta­tis­ti­cal analy­sis is not part of it.

Related Posts:

• Bench­marks for forecasting
• Fore­cast­ing work­shop: Switzer­land, June 2011
• Are we get­ting bet­ter at forecasting?
• Help for fore­cast­ing practitioners
• Mea­sur­ing time series characteristics