Using exploratory data analysis modified Box Plots to enhance Monte Carlo simulated Range Estimating Decision Technology

Abstract

Range Estimating Decision Technology (REDT) is a statistical analysis program designed to evaluate the quality of estimates. REDT provides two analysis of an estimate. 1) It evaluates the probability of attaining the estimated cost of a project. 2) It evaluates the line items of an estimate to determine which items have the greatest influence on potential gains or losses. However, current REDT cannot efficiently evaluate a data set as a whole and compare it to other sets of data. If this could be done REDT could potentially benefit in two ways. 1) An additional data analysis technique could help to further clarify an estimate improving its quality. 2) An additional analysis technique could provide additional information necessary for creating new applications of REDT. The Box Plot is a viable solution to evaluating data sets as a whole. It evaluates distributions and variabilities simultaneously and compares these measures of the data simply and understandably. The Box Plot works in unison with the other REDT analysis methods to measure exactly how much any given critical line item influences a simulations distribution of possible outcomes and its variability and how that effects the optimum determinate value. The subject of this thesis is to determine if an estimate can be substantially improved by using these measures of the data to modify an estimate. Two simulations of a hypothetical estimate were compared to evaluate the benefits of evaluating an estimate using the Box Plot analysis method. The Box Plot performed as expected, however the cost benefits for this detailed estimate were marginal. It was concluded that while the Box Plot analysis method does not dramatically improve an estimate, it does provide a means of supporting the reliability of current REDT analysis concepts and techniques. Additionally, the Box Plot provides two benefits to REDT. 1) It provides a simple but accurate means of comparing estimates. 2) It provides a means of determining what influence critical line items have on the overall distributions of estimates. The Box Plot may also have the potential to open doors to new applications of REDT in the fields of value engineering and competitive bidding.