Guestimator 101
Introduction
This article provides some further detail about Guestimator 101
How Guestimator 101 Estimates Defect Arrival Rates
For purposes of illustration, let's assume that Guestimator 101 uses the following very simple model for defect arrival rate. In practice, many more factors would have to be taken into account.
Guestimator 101 models the total number of defects created during the project by assuming a constant defect density, that is, the total number of defects is equal to the total number of lines of code times the defect density. Guestimator 101 has computed the defect density for each of the completed projects in its historical database, and has found that the values range from a low of 7 defects/KLOC to a high of 10 defects/KLOC, with 8 defects/KLOC being the most likely value. Since the estimate size of the project is 50,000 ESLOC, this gives the following range for the total number of defects:
Table of Estimated Defect Density and Total Defects
|
Density |
Total |
Low |
7/KLOC |
350 defects |
Most Likey |
8/KLOC |
400 defects |
High |
10/KLOC |
500 defects |
Guestimator 101 models the defect arrival rate as funtion of elapsed project time using a
Rayleigh distribution. This distribution is characterized by one parameter, T, that specifies the time at which the peak arrival rate occurs. Since the schedule for Tsunami 1.0 is 6 months, Syd uses the following value:
T = 2.5 months
For a given total number of defects, D, the defect arrival rate at time t is given by the formula:
defects/month(t) = D * (t / T^2) * exp(-t^2 /(2 * T^2))
Guestimator 101 uses this model to compute the low, most likely, and high estimates for the total number of defects that should be discovered each month. The following table summarizes this estimate:
Table of Estimated Total Defects Per Month
Month |
Low |
Most Likely |
High |
1 |
27 |
31 |
38 |
2 |
69 |
79 |
98 |
3 |
84 |
96 |
120 |
4 |
73 |
83 |
104 |
5 |
50 |
57 |
71 |
6 |
28 |
32 |
40 |
7 |
13 |
15 |
18 |
8 |
5 |
6 |
7 |
9 |
2 |
2 |
2 |
Comments
Enter your comments here:
Topic revision: r1 - 08 Jan 2010 - 21:58:39 -
ArthurRyman