STATISTICAL TECHNIQUES IN SOFTWARE SIX SIGMA AND DESIGN FOR SIX SIGMA (DFSS) - Software Design for Six-Sigma: A Roadmap for Excellence - page 130

TABLE 6.4 CPU Usage (in %)

55 52 55 52 50 55 52 49 55 52

48 45 42 39 36 48 45 48 48 45

65 62 59 56 53 50 47 44 41 38

49 46 43 40 37 34 31 28 25 22

64 61 64 61 64 64 61 64 64 61

63 60 63 58 63 63 60 66 63 63

60 57 54 51 60 44 41 60 63 50

65 62 65 62 65 65 62 65 66 65

46 43 46 43 46 46 43 46 63 46

56 53 56 53 56 56 53 56 60 66

of occurrence. The probability density function (pdf) curve can be constructed and
added to the graph by connecting the centers of data intervals. Histograms help in
selecting the proper distribution that represents simulation data. Figure 6.2 shows
the histogram and normal curve of the data in Table 6.4 as obtained from Minitab
(Minitab Inc., PA, USA). Figure 6.4 also displays some useful statistics about the cen-
tral tendency , skewness , dispersion ( variation) , and distribution fitness to normality .
Several other types of graphical representation can be used to summarize and
represent the distribution of a certain variable. For example, Figures 6.3 and 6.4 show
another two types of graphical representation of the yield requirement design output
using the box plot and dot plot, respectively.

Summary for Usage (%)

Anderson-Darling Normality Test

A-Squared 1.85

P-Value < 0.005

Mean 53.060

StDev 10.111

Variance 102.239

Skewness -0.766189

Kurtosis 0.171504

N 100

Minimum 22.000

1st Quartile 46.000

Median 55.000

3rd Quartile 62.000

24 32 40 48 56 64

Maximum 66.000

95% Confidence Interval for Mean

51.054 55.066

95% Confidence Interval for Median

51.742 57.258

95% Confidence Interval for StDev

95% Confidence Intervals

8.878 11.746

Mean

Median

51 52 53 54 55 56 57

FIGURE 6.2 Histogram and normal curve of data in Table 6.4.