Markov Decision Process can be used to evaluate a policy for repeatable situations. If you have some states that can occur repeatedly with some probabilities, then the Markov Decision Process can be used to evaluate the right action to take in a specific situation.
Markov Decision Process from SpiceLogic offers a very rich modeling application. It starts with a wizard that captures all the necessary information from you to create the model.
Once the wizard completes the data collection, it prepares the sophisticated Markov Decision Model Graphical User Interface, so that you can fine tune and optimize the model. The Same UI displays the evaluated policy as you can see here.
SpiceLogic Markov Decision Process offers a rich wizard for identifying objectives that can be used to attribute your rewards in a state. The reward can be modeled with multiple criteria.
And use Multiple Criteria attributes to model the reward of a state.
Using Analytic Hierarchy Process priority algorithms, your multiple objectives can be prioritized and quantified using pairwise comparisons. Even Transitivity rule can be enforced. Consistency ratio is also calculated and displayed while you do the pairwise comparison.
The state reward can be modeled with a sophisticated Utility function as well.
If you are not sure about your utility function, play the built-in Certainty Equivalent based Game to let the application infer your utility function. The inference logic is driven by behavioral economics.
A reward in a state can be uncertain. The uncertainty can be modeled with a probability distribution.
When a policy is generated by the Markov Decision Process, based on the recommended action, a Markov Chain is also generated. The generated Markov chain can be used to predict the future state after certain iteration.
You can even use custom expressions using the state names to predict a custom state of your Markov chain.
A lot of charts are available for you to get various perspectives of the Markov decision process.
Windows 7, Windows 8-8.1, Windows 10
4.5 or later