A guideline to statistical approaches in computer performance evaluation studies
Aridaman K. Jain · ACM SIGMETRICS Performance Evaluation Review · 1978
REVIEW 3.3.44.structure of data through models.• Estimation of parameters.• Examination of the adequacy of fitted models through residuals. Interpretation and Validation of ModelSummarization of the results.• Insight into the anticipated as well as the unanticipated.• validation of the model if it is to be used for prediction.• Relating back to the objectives.STATISTICAL BACKGROUND FOR USE BY CPE ANALYSTS 4.1 Methods of Tabulation and Display 4.2 4.3 4.4Histograms and stem-and-leaf displays.Summary Values Measures of central tendency.Measures of spread. Transformation, if necessary Analysis of VarianceA technique for partitioning the total variability into orthogonal and useful components.5.Testing of the hypothesis of the equality of means. RegressionDescription of one variable as a function of other variables.Interpretation of regression coefficients. Robust regression.Testing of Hypotheses. ClusteringMethods of clustering.References to examples of clustering.SUGGESTIONS FOR FUTURE WORK• Quantification of objectives.• Further work on representation of load on computer systems.Study of growth of load over time.More use of design of experiments and data analysis techniques.6.