Note: Currently in Beta. Feedback wanted and needed :)
The Interactive feature allows key values in a model to be altered and the forecast results compared. The intention is that by interactively working with these key values quicker collaboration and feedback can be found when forecasting proposed projects.
Exposing key model parameters
Parameters eposed at the top of model files become controls in the interactive window. Currently two types of parameters are available -
1. Date types: <?parameter name="StartDate" value="01-Jan-2016" type="date" ?>
2. Number types: <?parameter name="Developers" value="2" type="number" lowest="1" highest="10" step="1" ?>
Key feature of these parameters is the addition of the type="date" or "number" attribute. Some types, like "number" have additional attributes that customize the lowest value, highest value and the increment step size.
These parameters are used in the model just like other parameters by adding a @[param name) in place of a real value. For example:
<forecastDate startDate="@StartDate" ... >
<column id="3" estimateLowBound="2" estimateHighBound="3" wipLimit="= @Developers * 2" displayWidth="2">Develop</column>
Opening the interactive window
The interactive experiment window is opened by clicking the "Interactive" button in the "Home" ribbon tab. (This example is from the Kanban example - 2 - Typical Board)
Experiment trials are performed by altering the values to suit, and then clicking the "Run Forecast trial" button. The results and comparisons are shown in the lower portion of this dialog box.
1. Set the number of Monte Carlo cycles to 100 or 250 so the results are more stable (don't vary without the value inputs changing. Set this in the options area of the Home tab before launching the interactive window)
2. The forecasts are taken at the first level above the targetLikelihood value in the forecast setup. This is 85th percentile by default.
3. The compare tab has drop-down combo boxes to choose what trials you are comparing. These can help quickly find the fastest, lowest cost, or shortest duration trial settings.