Understanding
Software Estimating Models
This two day course is designed to provide you
with an advanced mathematics-oriented understanding of
currently-available software estimating models and
products.
Who Should Attend
Anyone interested in the mathematical basis of
software estimating models and products, the potential
benefits being:
- More effective use of currently-owned models and
products
- Better understanding of the basis of estimates
being provided by bidders and subcontractors
- Ability to make more informed buying decisions
when in the market for estimating models and
products
Purpose of the Workshop
- Review the key elements of the software project
management process and how they relate to the
software development process
- Review the core software project management
metrics and the fundamental laws of software project
dynamics
- Consider a categorization of software estimating
models by their mathematical forms
- Understand the mathematical derivation of the
models in each category
- Understand the behavioral differences between
models and how to quantify those differences
- Understand the stochastic nature of these models
and the proper random variable treatment of the
independent and dependent variables
What You Will Learn
Measurement
Objectifies Management
- Software development is a process
- Measurable process
è
predictable outcome
- Time, effort, defects
ç volume
AND efficiency AND defect vulnerability
- Estimating: judgment versus calculation
Models Can Be
Categorized by their Math
- Type 0 — dart board, dice, etc.
- Type 0.5 — engineering judgment
- Type 1 — single univariate power function
- Type 2 — two univariate power functions
- Type 3 — single bivariate power function
- Type 4 — bivariate 3rd-order polynomial
Models Have Unique
Behavioral Characteristics
- Entropy
- Economy
- Tradeoff sensitivity
Estimating is
Probabilistic
- Independent variables as random variables
- Selecting appropriate distributions to represent
the independent variables
- Using relevant probability theorems and Monte
Carlo methods for combining and convolving the
independent random variables
- Using geometric projection to solve for the
correlated dependent random variables
|
