The Markov Data add-in constructs tables that describe data for instances of specific application areas of Markov analysis. The Markov Data add-in calls the Markov Models add-in to construct the model form to hold the stochastic features of the model. The Markov Data add-in then fills the model form with parameters linked to the data form. The resulting form will ultimately be used to provide data for one of the solution add-ins.

Although the model forms available in the Markov Models add-in are very general, it is somewhat difficult for the unskilled user of Excel to build a model. The Markov Data add-in removes this difficulty by automatically building the model for specific problem classes. At this time only three classes are available, the birth-death process, the finite queue, and the random walk. For each class there are several variations. Future editions of the add-in may include more problem classes.

The addition of new problem classes requires VBA programming. The programs of this add-in are not password protected so you can examine the programs for existing classes. As a user of this add-in, you may find it interesting to add new classes. Several program features make this not too difficult. I am glad to accept contributions of new classes and will include them in future editions. I will also be willing to help aspiring contributors.

The picture to the left shows the menus of the Markov Data and Markov Models add-ins. To use the Markov Data add-in select the Data item from the menu.

The Start and Finish commands are used to respectively add and remove buttons from the worksheets. This is important when moving models from one computer to another. Always delete the buttons with the Finish command before attempting to open a model on another computer. Otherwise you will experience an error message regarding links. Add buttons back with the Start command.

Selecting the Data item will present the dialog at the left. The problem type is selected with a button at the left of the dialog.

The dialog provides three model types. the Discrete Time Markov Chain (DTMC) considers time in discrete steps and specifies probabilities of transition. The Continuous Time Markov Chain (CTMC) considers time as continuous and specifies rates of transition. The Markov Decision Process (MDP) describes a DTMC but adds decisions to the model. This feature requires a more complex model, but many interesting decision problems may be considered. All of these model types are considered by the other add-ins in this collection, but here we describe only the data portion. The Model Type is selected by clicking the appropriate button in the Model portion of the dialog.

To analyze DTMC or CTMC models the Markov Analysis add-in must also be installed. The analysis of MDP models requires the Markov Decision add-in.

Clicking the OK button presents a second dialog with data specific to the problem type. The Model Name field accepts a name and is initially filled by a default name. The name may be changed on this dialog, but not after the model is constructed. If you choose a new name, make it small, include no spaces, start the name with a letter, and do not use punctuation. The names "Prob 1" or "Prob.1" will both fail because the first contains a space and the second a period.

The maximum population determines the size of the model and the corresponding solution effort. Although the add-ins do not restrict the size of the problem entered, some computational methods are impractical for large problems. Analysis methods use tables with the number of columns proportional to the number of states. The number of columns is limited by Excel 2003 and 2004 to 256, thus limiting the size that can be analyzed. Excel 2007 has a much larger column capacity. I have not tested the methods for large problems. Some computations in the Markov Analysis add-in use the Excel INVERSE function that is limited to about 50 states. The size of a model, as measured by the number of states, is an important feature of a Markov model because for many problems the number of states is very large.

The remainder of this section describes the problem types and their models. For more detailed discussions of the modeling and solutions processes, see the other parts of this Markov collection.