Estimation and Measurement Telecon, 2009-07-31

See Weekly Meeting Logistics for telecon information.

Attendees

AndyBerner, ArthurRyman, LawrencePutnamJr, LeeFischman, SteveAbrams

Regrets

ScottBosworth

Minutes

1. Review discussion of How to represent effort by date - AndyBerner

AndyBerner led the review of how to represent effort.

Labor Categories and Activities as Dimensions

LeeFischman described the SEER approach which is to break down effort by labor category, e.g. Management, Test, Code, and activity, e.g. Architecture, Design, Implementation. SEER has 8 labor categories and 8 activities which gives an 8 x 8 matrix of effort by labor and activity. Furthermore, this matrix is associated with a given confidence level, e.g. 50% probability. Since this matrix can be estimated at any desired probability, we could do this for a set of quantiles. The matrices therefore can be combined into a single matrix whose entries are quantile functions for effort. These quantile functions can be summed to give the effort by labor, the effort by activity, or the total effort.

LawrencePutnamJr described the QSM approach which is to provide effort by labor category and labor by activity. The more detailed 2-dimensional data can be produced by applying weighting factors for each labor category per activity.

ArthurRyman remarked that this is a special case of dimensional data which is used in analytical processing. In this case we have two dimensions: labor category and activity. The measure is effort, with the added complexity that the measurement is a quantile function. We will have a general XML format for representing dimensional data.

[ACTION] ArthurRyman to specify the XML format for dimensional data.

ArthurRyman requested same realistic sample data that included the labor categories, activities and efforts.

[ACTION] LeeFischman to attach a SEER example to MetricsEffortEstimateDetail.

[ACTION] LawrencePutnamJr to attach a SLIM example to MetricsEffortEstimateDetail.

Properties That Affect Effort Estimates

LeeFischman described several properties, e.g. the number of productive hours in a day or the number of work days in a month, that, in his experience, have had a significant effect on effort estimates. These should be standardized for use in the EMS 1.0 interface.

Mapping Versus Standard Terms to Achieve Interoperability

AndyBerner raised the issue that we will need a way to resolve differences in terms used by each vendor, e.g. for labour categories.

SteveAbrams said that one way was to define standard OSLC terms.

LeeFischman said there was other ways to handle this.

Due to the importance of this issue and the lack of time remaining, we agreed to defer the discussion for two weeks. The discussion will be a faux debate between SteveAbrams and LeeFischman. Each person should post a position statement in advance.

[ACTION] SteveAbrams to post position statement by 2009-08-13 at MetricsMappingVersusStandardTerms

[ACTION] LeeFischman to post position statement by 2009-08-13 at MetricsMappingVersusStandardTerms

2. Review working draft of EMS 1.0 Part 1: Resources and Services - ArthurRyman

ArthurRyman reviewed the proposed change in architectural viewpoint, i.e. that all tools are consumers of the EMS 1.0 service. There was a lot of discussion on this topic. Due to the insufficient time remaining, people should review the write-up and continue the discussion next week.

3. Other Business - All

AndyBerner will be on vacation next week and miss the telecon.