There are several validation methods for software cost estimation using Machine Learning models. But, I was wondering if there are any validation methods for Algorithmic and Non-ALgorithmic models such as COCOMO, Function point, analogy, Putnam, etc.
Is there any paper that could guide me related to this question?
Hi Sai,
This book may be beneficial:
https://www.barnesandnoble.com/w/software-engineering-economics-barry-w-boehm/1101636896?ean=9780138221225
-Adrian
If you mean validation of the results , you can measure the effort estimated in working hours then use the R square to find the difference.
please see my research thesis and the paper , the thesis materials are available online.
it is very helpful for you in this regards. where my work was in Function point measure
see the links :
Thesis NEW COMPLEXITY WEIGHTS FOR FUNCTION POINT ANALYSIS USING ART...
Article Modification of standard Function Point complexity weights system
regards
You are of course assuming that cost (and effort) models are validated, right!?
I had a student a number of years ago who was looking in to the validity of effort estimation from multiple organisation data sets. That might be worth a read to give you an idea of some of the issues:
https://aut.researchgateway.ac.nz/bitstream/handle/10292/473/DengK.pdf
I believe for estimating the cost estimation models, there are numerous validation/accuracy evaluation techniques such as Mean of Relative Errors (MRE), Pred (0.25) and the median magnitude of relative error (MDMA) are the most used by the researchers in the field.
Accuracy Comparison of Analogy-Based Software Development Effort Estimation Techniques by Ali Idri might help you to understand.
Following readings may help:
1. M. Shepperd and C. Schofield, "Estimating software project effort using analogies," IEEE Transactions on Software Engineering, vol.13, pp. 736-743, 1997.
2. S.D Conte, HE Dunsmore, and VY Shen, Software Engineering Metrics and Models, Benjamin-Cummings, Menlo Park CA, 1986.
- Badges
- Science topic