Importing Stochastic Element Definitions from a Spreadsheet

You can use the Spreadsheet element to import a probability distribution from a spreadsheet. The distribution is defined in the spreadsheet using a number of fields that specify the type of distribution, as well as the arguments to the distribution.

Probability distributions must be specified in the spreadsheet in a very specific format. In particular, the distribution must be defined in a row of adjacent cells in the spreadsheet. The first cell is a code that defines the type of distribution. The subsequent cells (to the right of the code) define the input parameters for the distribution. The table below specifies how each type of distribution must be entered in the spreadsheet.

Distribution	Type Code	Distribution Parameters
Generalized Beta	BETAGEN	Mean	SD	Min	Max
Beta	BETASF	Successes	Failures
BetaPERT	BETAPERT	Min (0%) or 10%	Most Likely	Max (100%) or 90%	0 for 0/100; 1 for 10/90
Binomial	BINOM	Batch Size	Prob
Cumulative	CUM	# pairs	Linear = 0; Log = 1	Prob1	Val1
Discrete	DISCRETE	# pairs	Prob1	Val1	Prob2	Val2
Exponential	EXPON	Mean
Extreme Probability	EXTRPROB	Min = 0; Max =1	Number samples
Extreme Value	EXTRVAL	Min = 0; Max =1	Location	Scale
Gamma	GAMMA	Mean	SD	(Min)	(Max)
Log-Normal	LOGNORM	True = 0; Geom = 1	True or Geom Mean	True or Geom SD	(Min)	(Max)
Negative Binomial	NEGBINOM	Number successes	Prob of success
Normal	NORMAL	Mean	SD	(Min)	(Max)
Pareto	PARETO	Shape	Mode	(Max)
Pearson Type III	PEARSON	Location	Scale	Shape
Poisson	POISSON	Expected Value
Sampled Results	SAMPLED	# Results	Val1	Val2	Val3	Val4
Student’s t	STUDENTT	Deg of Freedom
Triangular	TRIANG	Linear =0; Log =1	Min (0%) or 10%	Most Likely	Max (100%) or 90%	0 for 0/100; 1 for 10/90
Uniform	UNIFORM	Linear =0; Log =1	Min	Max
Weibull	WEIBULL	Min	Slope	Mean – Min	(Max)

• If Log-Normal distribution is defined using Geometric parameters, the SD is dimensionless

• Shaded parameters are dimensionless; others have same dimension as the distribution

To import a distribution from a spreadsheet, you must press the Type button in the Import dialog:

After doing so, the following dialog will be displayed. You must select “Probability Distribution” as the Type:

When you specify the Location, you must select a single cell: that corresponding to the Type Code for the distribution. GoldSim will automatically read in the appropriate number of additional cells based on the distribution type.

Note: Prior to running the model, you can view any spreadsheet data that you plan to import by first selecting the Update Spreadsheet Outputs item in the menu displayed via the Options>> button for the Spreadsheet element, and then viewing the outputs of the element in tool-tips within the element’s output port.

Note: Care must be taken when importing distributions that represent dates and temperatures. This is because dates and temperatures both have an absolute reference (i.e., 12/31/1899 for dates and absolute zero for temperatures). Differences between dates are expressed in units of time (e.g. days, seconds, etc) and differences between temperatures are expressed in ‘deg’ units (Cdeg, Fdeg). This can lead to errors when importing Stochastics, since GoldSim assumes a single unit for the distribution parameters, while some types of distributions would require two different types of units. For example, a Normal distribution representing a temperature would require the Mean to be specified in absolute units (e.g., C) and the Standard Deviation to be specified in difference units (e.g., Cdeg). Similarly, a Normal distribution representing a date would require the Mean to be specified in absolute units (e.g., as a date) and the Standard Deviation to be specified in difference units (e.g., days). If faced with this problem, there are several approaches for addressing this: 1) Use a distribution type where all parameters have the same units (e.g., triangular, uniform, cumulative); or 2) Specify and import the distributions as dimensionless and then apply a unit to the sampled value using an Expression element. This latter option would work for temperatures, but not dates, since a date cannot be specified with dimensionless values. It could, however, be imported as a Julian date, in which case you could use day units (or some other time unit). Alternatively, the mean date could be defined as a Data element and an imported distribution with time units could be used to model a ‘lag’ (positive or negative) from the fixed date. An Expression element could be used to add the sampled lag value to the fixed (mean) date.

When a distribution is imported in this way, the output type on the spreadsheet element is not a value; rather, it is a complex output referred to as a Distribution output. Distribution outputs are complex outputs that represent all the statistical information necessary to define a probability distribution and only be used in several specialized locations.

To import this information into a Stochastic element, you must then use this Distribution output as the definition for a Stochastic element specified as an Externally-Defined distribution.