Data and System Models in Predictive Systems Lab

Top  Previous  Next

 

In Predictive Systems Lab, data models are used to organize document data into a set of independent and dependent variables for the purpose of approximating the behavior of some physical event. Predictive Systems Lab can organize the data whether you are required to estimate the state of a physical system under specific conditions, or predict future states of a time-evolving system.

 

Physical models, on the other hand, are viewed by Predictive Systems Lab as black boxes where the nature of the physical system is of no relevance, but the information that goes into the system and the information that comes out of the system is important instead. The system itself is left as an unknown function to be generalized.

 

BlackBoxFig

 

In this context, Predictive Systems Lab equates inputs to a system as independent variables, black boxes as unknown functions, and the outputs of a system as the dependent variables. When specifying a data model, you are required to supply inputs and outputs, as opposed to independent and dependent variables.

 

Three classes of systems can be generalized in Predictive Systems Lab:

 

1.Systems that depend on a set of specific conditions.
2.Systems that depend on their own past states.
3.Systems that depend on both specific conditions and past states.

 

Predictive Systems Lab makes no particular distinction among systems you define. It does apply restrictions as to what can be inferred from a generalized model. Systems that depend on specific conditions are not possible to foretell in time, but can be approximated for a given set of conditions. Systems that solely depend on their past can be forecasted, at least theoretically, ad infinitum in time.

 

Current software limitations on the number of rows and columns that can be contained in a Predictive Systems Lab document worksheet (see Specifying Cell Ranges and Locations for a discussion of this subject), have confined the designation of variables to data arranged in columns. As such, any one system can be defined to a maximum of 8,192 variables, the maximum number of columns, with a maximum of 4,194,303 values for each variable, the maximum number of rows. Therefore, in Predictive Systems Lab systems, columns are regarded as variables, and rows as the snapshot of these variables in time.

 

NoteAny one of the possible 8,192 variables can be defined as a time-delayed vector; therefore, the maximum possible number of values entering the system at any given time can be many times that limit.

 

TipYou can approximate a model with as many methods as you like, but the current implementation of Predictive Systems Lab only allows one model per document worksheet. However, you can add up to 255 sheets to any document and in this fashion define an equal number of models.        

 

Sponsored

Try Predictive Systems Lab

Interactive Demo →

Need Help?
Contact support