Specifying What the Input to a Time Series Represents

The input to a Time Series element is always a set of time series records (i.e., a table consisting of time/value pairs). However, since a time series can represent several different kinds of data (e.g., values, rates), the first (and most important) step in defining a Time Series element (prior to actually defining the records) is to specify what these records represent. GoldSim provides several different ways for you to define the inputs to Time Series elements.

In particular, you can specify the input to a Time Series as representing one of six kinds of data:

• The instantaneous value of a variable at specified times: Examples include the price of a commodity (e.g., gold), the price of an asset (e.g., a stock), or the height of a projectile.

• The constant value of a variable over the next time interval or the constant value of a variable over the previous time interval. In some cases, a value is assumed to be constant over a specified interval (i.e., follow a "stair-step" function from interval to interval). In some instances, this may reflect the fact that the variable actually is held constant over each interval. More commonly, it may be an artifact of how the data was recorded (e.g., it may have only been recorded in terms of an average value over the interval). GoldSim provides two options for entering this kind of data (one is forward-looking, and one is backward-looking).

• The (uniform) change (increment) in a variable over the next time interval or the (uniform) change (increment) in a variable over the previous time interval: These options are usually used to output cumulative values (e.g., cumulative sales, cumulative rainfall) and/or rates (e.g., sales rates, rainfall rates) that are specified by defining incremental changes in a variable (e.g., monthly sales, monthly rainfall) over specified intervals. When this option is selected, the Primary Output of the Time Series is not the specified time series itself; rather, it is the cumulative value of the variable at each point in the simulation (assuming that the specified changes occur uniformly over each interval).

Good examples of this type of time series data are rainfall and sales rates. In these cases, what is actually measured and recorded are totals (e.g., rainfall in mm or sales in $) over a particular interval (e.g., 31 days). Based on this data, you could then use GoldSim to compute the cumulative rainfall and sales (over the entire simulation), as well as the average daily rainfall or sales rate for each interval (e.g., month). GoldSim assumes that the variable (e.g., cumulative rainfall or sales) changes uniformly over the time period, which means that the rate of change is constant over the interval (i.e., it "stair-steps"). GoldSim provides two options for entering this kind of data (one is forward-looking, and one is backward-looking).

• Discrete changes to a variable at specified times. In some cases, your time series may not represent the current value of a variable, but a series of discrete changes (additions and subtractions) to a variable. An example of this kind of time series deposits and withdrawals from an account that happen discretely. Note that this option is only available if the Data Type has been defined as a Discrete Change Signal with an Add instruction.

Prior to defining the time series records, you must select one of these four options from the drop-list labeled Represents.

After specifying what the input to the Time Series represents, the next step is to define the actual Time Series input data.