November 25, 2003
Estimating
Estimating is an art.
It's not like coding where one can write a unit test in JUnit, click a button and see whether the bar shows green or red. Estimating is obviously not cut and dried like that. For one thing it involves predicting the future. How many times, despite modern technology, a wealth of knowledge and experience, do weather forecasters get it wrong? Plenty.
So estimating isn't easy. But sensible guidelines can help. I am indebted to my colleague Alan for his method. I particularly like the emphasis on assigning a risk to each task and an associated schedule allowance or risk factor as I like to call it.
Obviously the accuracy of any estimate is dependent on the knowledge of the task and there will be more unknown aspects to some tasks than others. And for other tasks we will be able to identify associated risks. So it makes sense to weight the estimate for each task according to its risk.
One final thought. As with any art, estimating ability is likely to improve with practice.
Posted to Software Development by Keith Pitty