Tuning of Cost Drivers by Significance Occurrences and Their Calibration with Novel Software Effort Estimation Method

Algorithm 1

Sub-Model  1:

Step  1. Generate the MMRE (M) for Available 𝑁 Projects using actual and COCOMO estimated efforts.

(i) [BEGIN]

(ii)  Input the 15 cost drivers, KLOC, Actual Effort for NASA projects.

(iii) [LOOP]

  for 𝑗=1 to no. of projects (say 𝑛)

  EAF[𝑗] = D1 * D2 *⋯* D15

  Estimated Effort[𝑗] = 𝑎[𝑗] * (kloc[𝑗]∧ 𝑏[𝑗]) * EAF[𝑗]

  MRE[𝑗] = ActualEffort[𝑗]−estimatedEffort[𝑗]/Actual Effort[𝑗]

  MMRE (original) += MRE[𝑗]

  MMRE (original) /= 𝑛 [The original MMRE is obtained and noted down]

(iv) [END OF LOOP]

(v)   [END]

Sub-Model 2:

Step  2. for 𝐼=1 to 15

 temp = Emi

 Set Emi = 1

 Calculate Influenced MMRE(MN)

 List[𝐼,1] = 𝐼;

 List[𝐼,2] = MN~M;

 Emi = temp;

 end for

Sub-Model 3:

Step  3. Sort the list according to the second parameter in descending order

 For 𝐼=1 to 14

 For 𝑗=2 to 15

 If (list[𝐼,2] < list[𝑗,2])

 then

 swap (list[𝐼], list[𝑗])

 end if

 end for

 end for

Step  4. Sig = list[𝐼,1] represent the order of Significance occurrences.

Sub-Model 4:

Step  5. for 𝐼=1 to 15

 for 𝑗 = very low to Extra high (Six rating of cost driver)

 Select Projects (P) as an input for calculating the fitness value using fitness function F1 = MMRE(P).

 Set the range R as {Rmax, Rmin}

 Generate initial population for the cost driver with Range R.

 performs The Genetic operations for K generations.

 (1) Tournament Selection

 (2) Crossover with Pc = 0.8

 (3) Mutation with Pm = 0.3

 Select the individual (CDNEW) with the best MMRE

Step  6. Calculate the MMRE(Mmod) by replacing CDNEW with CDij

 if (Mmod < M)

 then update the value of CDij and M.

 else

 discard the value

 end if

 end for

 end for