The Nesma perspective on the current state of the art in the automation of functional size measurement for the specialized workshop on the IWSM in Gothenburg in Sweden on October 24, 2017.
2. Automation is a hot topic in the market
There is a market demand for timely information about project value
With Agile and DevOps the decision moments are more frequent
Manual size measurement can only fulfill these market demands in part
documentation must be sufficient for rapid sizing and only functional size
Project value / project cost is more than functional size
Several Nesma-related initiatives in this area:
Estimating total project cost by sizing experts and architects (IWSM paper)
CAST user initiative to match manual and automated functional size
| FSM Automation | State of the art from a Nesma perspective
3. Recent contributions of Nesma members
October 2017 - IWSM/Mensura conference
A Shortcut to Estimating Non-Functional Requirements
An approach to use Architecture Driven Estimation mechanisms that include more than
functional size to create good cost predictions.
October 2016 - Nesma autumn conference
AFSFPA (Automated Functional Sizing for FPA)
A tool to determine functional size in Nesma FPA with up to 90% accuracy from design
artefacts
May 2016 - ICSSP conference
An Exploratory Study on FSM based on code
An exploration of opportunities, challenges and obstacles of FSM based on code as
perceived by 336 FSM specialists from around the world
| FSM Automation | State of the art from a Nesma perspective
4. Automated FPA versus manual FPA?
A growing number of companies measure their full software lifecycle
with function points:
Investment decision Planning Payment for
delivered software
Three different objectives, three different methods. Is this a problem?
| FSM Automation | State of the art from a Nesma perspective
Indicative FPA on
early specifications
(manual/workshop)
High-level FPA on
final specifications
(usually manual)
High-level FPA on
delivered software
(pref. automated)≠ ≠
5. The real problem: Size versus Money
Investment decision Planning Payment for
delivered software
| FSM Automation | State of the art from a Nesma perspective
10,000
100,000
1,000,000 or more?
FP * Productivity FP > Money
≠ ≠NFR
Non-Functionals
=
Time & Material
6. Automation challenges to address
Bringing automated FSM from experiments to industry
All Software Measurement Associations should promote automation
Creating etalon specifications for automated FSM from requirements
based on publicly available specification documents
Creating a shared vision on how to compare different approaches
Is it possible to have etalon software with the etalon specifications
Creating a standardized vision how to deal with NFR
This vision should translate to rules for automation
Different NFR approaches should be comparable somehow
| FSM Automation | State of the art from a Nesma perspective
7. Automation challenge #1
Bringing automated FSM from experiments to industry
Publish known use of automated FSM in industry
- Renault uses automated FSM from detailed specifications
- CAST has a number of industrial clients
- . . . .
Publish
Scale up
| FSM Automation | State of the art from a Nesma perspective
8. Automation challenge #2
Creating etalon specifications for automated FSM
To independently verify accuracy of different tools
Public specifications available from government projects or GitHub
Create
Community
| FSM Automation | State of the art from a Nesma perspective
9. Automation challenge #3
Creating a shared vision on how to compare different approaches
Investment decision Planning Payment for
delivered software
Vision
| FSM Automation | State of the art from a Nesma perspective
Indicative FPA on
early specifications
(manual/workshop)
High-level FPA on
final specifications
(usually manual)
High-level FPA on
delivered software
(pref. automated)≠ ≠
10. Automation challenge #4
Creating etalon software for automated FSM
To independently verify accuracy of different tools
To compare tools that measure specifications and code
Public specifications available from government projects or GitHub
Create
Community
| FSM Automation | State of the art from a Nesma perspective
11. Automation challenge #5
Creating a standardized vision how to deal with NFR
This vision should translate to rules for automation
Different vendors measure different aspects
CAST measures AEP, a number that includes Functional Size & NFR
IFPUG uses SNAP-points on top of FP. Can they be automated?
SIG uses a five-star scale, to indicate maintainability
TIOBE uses a A-G rating, to indicate software quality
. . . . .
Can the proposed Nesma framework help?
| FSM Automation | State of the art from a Nesma perspective
Standardize
12. There is work to do
| FSM Automation | State of the art from a Nesma perspective