It's a number. Just like the average speed on your sports tracker. You don't estimate it. You know it based on real data. Not some fuzzy ritual.
This is yet another uninformed notion. The average is the calculated central value of a set of numbers. The mean of a series of numbers. This is the arithmetic mean of that series. There are other means, but for software estimating the mean is the correct term.
However that mean is NOT just a number, it's a calculated number from the series. And that series has other useful statistics. The Variance for example which is the expectation of the squared deviation of a random variable from its mean. Variance measures how far a set of (random) numbers are spread out from their average (mean) value. The variance is the square of the standard deviation, the second central moment of a distribution, and the covariance of the random variable with itself.
Notice here the term random variable. All the variables in software development are random variables. That notion in the quote based on real data is actually based on the random variables sampled from the past project's performance.
The underlying stochastic processes of software development creates random variables in accordance with the processes generating the random variable - a generating function. These random variables can be collected from past performance OR they can be generated in a simulation process that models the underlying stochastic processes of software development.
Monte Carlo Simulation and Method of Moments are processes used to model the cost, schedule, and technical performance of projects.
That real data mentioned in the quote is random real data, so modeling how that random data will impact the future requires several steps
- Are there a sufficient number of samples of past data to produce a statistically credible model of the future. There are several tests for this.?
- Is the underlying stochastic process generating these random values stationary - that is will it be the same in the future as it was in the past?
- Are the uncertainties that drive the stochastic processes understood, so the proper use of the real data and any modeled data be assured? There are tests for this as well.
The Fallacy of the Week
The fallacy is that the number for the mean is a random number, not a cardinal number. That number is subject to change in the future, driven by the underlying uncertainties of the process. So know the mean (average) requires you know the Variance (square of the Standard Deviation), the stability of the generating process, and several other statistics before that number is of any real use for making decisions.
When you hear what was stated in the quote, it's an indication that the speaker is unaware of the statistical processes that drive the estimating processes, their modeling, and their use in decision making.
