PROMOTING ACCESS TO AFRICAN RESEARCH

Global Journal of Mathematical Sciences

Log in or Register to get access to full text downloads.

Remember me or Register



Estimating software development project size, using probabilistic techniques

Benjamin Oyediran Oyelami, Samuel Bisong Oyong

Abstract


This paper describes the quantitative process of managing the size of software development projects by Purchasers (Clients) and Vendors (Development Houses) where there are no historical databases. Probabilistic approach was used to estimate the software project size, using the data collected when we developed a tool to estimate the time to complete a new software project, and from The University of Calabar Computer Centre. The Expected size of the Tool was estimated to be 1.463 KSLOC, but when the tool was actually coded, it was found to have 1.530 KSLOC. To validate the tool further, the expected size of revenue collection program for Equity Bank Plc in Nigeria was estimated to be 17.91 KSLOC, but when the program was coded, the actual size was found to be 17.40 KSLOC. Using a two-tailed test it was found that these results were good estimates as the sizes fell within the 99.80 % confidence limit. That is:


P (μ - 3σ2 ≤ X ≤ μ + 3σ2) = 99.8%


Where

X = Expected size (KSLOC)

µ = 17.91 KSLOC= Average source lines of code (mean)

σ2 = 0.2326 KSLOC = Total standard deviation

Standard error = 3.09

P = probability

Level of significance α = 0.1

KEY WORDS: Expected size, Uncertainty, Size range, KSLOC, Vendors.

Global Journal of Mathematical Sciences Vol.3(2) 2004: 179-184



http://dx.doi.org/10.4314/gjmas.v3i2.21366
AJOL African Journals Online