Estimation and Measurement Telecon, 2009-10-02

See Weekly Meeting Logistics for telecon information.

Attendees

AndyBerner, AndrewCanham, ArthurRyman, LawrencePutnamJr, LeeFischman

Regrets

SteveAbrams

Minutes

1. Alternative XML Format for Fact Tables - SteveAbrams

SteveAbrams was unable to attend today. Defer this to next meeting, 2009-10-16.

2. Software Estimation Use Cases Status - LawrencePutnamJr

LawrencePutnamJr and AndyBerner held a telecon to complete the Reestimating use case. They reviewed and discussed the result:

• There are two main triggers for reestimation: 1) passage of time, e.g. a phase completes, and 2) project assumptions change
• For 1) the happy path is that the project is on track. The metrics will be reestimated to completion based on the actuals. This quantifies the reduced investment risk.
• For 1) the unhappy is that the project is not on track. In this case, the project assumptions need to be changed outside the normal allowed envelope of variation, e.g. reduce scope, extend schedule. New actors are involved: engineer, executive
• For 2) the project needs to be reestimated based on the new project assumptions.

ACTION: AndyBerner will update the description of the reestimation use case based on the above discussion.

3. Primer Status - ArthurRyman

ArthurRyman reviewed the status of the Primer:

• The Initiating scenario was getting very long so it has been split into three acts.
• The Primer shows labour rates being used in the portfolio management tool. LawrencePutnamJr confirmed that estimation tools also use labour rates. We should treat labour rates like other project assumptions.
• LawrencePutnamJr asked about Work Breakdown Structures (WBS). It was agreed that WBS was a valid piece of information to exchange, either as project assumptions or the result of an estimate. Initial thinking is to use Microsoft Project 2007 XML format since this is an available defacto standard and can be generated using any XML programming techniques, i.e. it does not require any code from Microsoft. It our experience with this is negative, we also have the option of defining a simpler OSLC format for use in EMS 1.0.
• We agreed to begin implementation when the Initiating use case was adequately specified, i.e. we will work iteratively.
• ArthurRyman raised the issue of how the probability distributions of effort by role roll up to total effort: are the estimates for each role statistically independent or are they correlated, e.g. by being some fixed portion of the total effort such as 2:1 coder to tester ratio?
• LawrencePutnamJr said that SLIM estimates the probability distribution for the total effort and applies a numerical factor to the total to obtain the effort for each role.
• ArthurRyman suggested that this is a useful simplification since less data needs to be exchanged and the relation between the probability distributions for the total and each role is well-defined.

4. Other Business

• Arthur will be be in transit on 2009-10-09. We will skip that meeting and resume 2009-10-16.

Comments

Add your comments here: