Models

The Repeated Measures Degradation platform has many built-in models that you can use in the Bayesian estimation procedure. The number of models that are available depends on if an X variable was specified in the launch window.

Note: The thumbnail sketch shown to the left of each equation shows a generic plot of the behavior of the path definition over time. In the report’s main plot, the plot of the estimated models can differ from the thumbnail based on your selections for Transformation.

Models in the Repeated Measures Degradation platform contain fixed parameters and random parameters. Both types of parameters are unknown. However, fixed parameters are the same across all values of the Label, System ID variable. Random parameters, on the other hand, vary across values of the Label, System ID variable.

After you choose a model, you then select distributions for each of the random parameters, assign prior distributions to the location and scale parameters and correlation coefficients for the random parameters, and assign prior distributions to the fixed parameters. For more information about the model specification, see Statistical Details for the Repeated Measures Degradation Platform.

If you do not specify an X variable in the launch window, there are 6 models available. These models are listed in Figure 8.9. The models appear in pairs, where the first of each pair having a fixed intercept and the second of each pair having a random intercept.

Figure 8.9 Models Without an X Variable 

When you specify an X variable in the launch window, additional models are available in the Path Definition panel. Figure 8.10 shows the first-order kinetics models. Figure 8.11 shows the constant rate models.

Note: The indented models are variations of the unindented models immediately preceding them. The variation is that the initial condition is shifted or that the initial conditions are random.

Figure 8.10 First-Order Kinetics Models 

Figure 8.11 Constant Rate Models