by Ed Merriman
| August 27, 2015
Original Post Date: July 1, 2015
Consumers of cost estimates need to understand how risk and uncertainty impacts their estimates and, as such, performing Monte Carlo analysis is becoming a required step in producing a cost estimate at many organizations. One of the most popular Monte Carlo tools is Oracle’s Crystal Ball and PRICE has produced an Excel based solution that allows users to perform Monte Carlo analysis on their TruePlanning estimates. With this combined solution users can run risk/uncertainty analysis on their parametric estimates using an industry standard tool.
In the TruePlanning/Crystal Ball solution, users select TruePlanning inputs to be used as “Assumptions” in Crystal Ball. Assumptions in Crystal Ball are the ‘inbound’ values that are varied over a curve with each iteration of the Monte Carlo run and drive the ‘outbound’ “Forecasts”. The forecasts are recorded and become the data set used to drive the final results of the Monte Carlo run. In previous versions of the TruePlanning/Crystal Ball solution when a TruePlanning project was loaded each Assumption was given a triangular distribution based on a fixed percentage of the initial input’s value. Users would then update each assumption’s setting in Excel. They could change the distribution, or simply change the values used in the triangular distribution.
A major problem with this was that if the TruePlanning project changed, the project needed to be reloaded into the TruePlanning/Crystal Ball solution and all the Assumptions that had been previously set by the user would be overwritten. On a large projects this represented a major effort.
To resolve this issue, the TruePlanning/Crystal Ball solution has been updated to use the Pessimistic and Optimistic risk/uncertainty values stored in TruePlanning for each input. There are many advantages to this enhancement including: the fact that users will no longer have to reset Assumptions in Crystal Ball when they reload projects; Assumption values will survive PBS changes; users can leverage the TruePlanning interface to set risk/uncertainty values easily, including using the ‘Find and Replace’ dialog; and users can use the same risk/uncertainty values in the TruePlanning provided FRISK analysis as they do in Crystal Ball.
If an input that has been selected for use in Crystal Ball does not have risk values set in TruePlanning, the values set in the Crystal Ball Assumption will be 50%, 100% and 150%. If risk values are supplied in TruePlanning, the ‘Calculated Pessimistic’ and ‘Calculated Optimistic’ values will be used. The pessimistic value will be used as the maximum value in the Crystal Ball triangular assumption, and the optimistic value will be used as the minimum value in the Crystal Ball triangular assumption. By using the ‘Calculated’ risk values from TruePlanning users can flow changes directly from the input values in TruePlanning to their Monte Carlo analysis.
Some TruePlanning inputs are ‘inverted’ in terms of risk. Inputs like ‘Weight of Structure’ or ‘New Code Size’ are inputs that increase cost as they get larger so the pessimistic value will be higher than the optimistic value. There are some inputs that decrease cost as they get larger. An example would be the “Organizational Productivity” input for Software Component. This input’s value gets larger to represent more efficient organizations, so as this input increases in value, the estimate it is used to calculated decreases. This needs to be reflected in the risk analysis. This means the optimistic value needs to be larger than the pessimistic value. The TruePlanning/Crystal Ball solution is aware of these inputs and will automatically invert the values used in the triangular Assumption in Crystal Ball.
To date, and with the new enhancement, the TruePlanning/Crystal Ball solution has been used to help PRICE’s customers produce and defend major, highly visible parametric cost estimates that include Monte Carlo risk analysis. To obtain a copy of the newly enhanced TruePlanning/Crystal Ball solution, please contact PRICE Systems.