Distribution Comparisons of EAC Cost Growth
The ATF is a method of equally trimming a percentile from the lowest and highest data in a data set (Bednar & Watt, 1984) to minimize the effect of very low or high numerical values on generally calculating the mean. This same logic can also be applied to the cumulative distribution function. Using this method, we remove the top and bottom 1% of each cost growth dataset prior to fitting a probability distribution. This results in truncating the distributions on both the left-hand and right-hand side of the distribution. Figure 2 shows the ATF performed on the data from Figure 1 and the more appropriate fitting of the Rayleigh distribution. The population median of this distribution now equals 1.9, a more accurate representation of the dataset’s median of 1.5. It is evident the ATF method results in fitting the data much closer to the actual data. Therefore, the ATF method is applied to all analysis for curve fitting in this article. Table 1 reflects general descriptive statistics of the calculated cost growths by WBS element to which we fit probability distributions. Table 2 continues these descriptive statistics by presenting the correlation coefficients (McClave et al., 2017) between the Level 2 WBS elements. Table 3 displays the 59 programs from which we calculated these cost growths.
FIGURE 2. RAYLEIGH FIT TO WBS COST ELEMENT WITH ATF
1.2
Histogram of Data Rayleigh Distribution
1.0
0.8
0.6
0.4
0.2
0.0
0 1
2
3
4 5
6
7
Note. ATF = Alpha Trimmed Filtering; WBS = Work Breakdown Structure.
38
Defense ARJ , Spring 2025, Vol. 32 No. 1
Made with FlippingBook - Online Brochure Maker