Scenario
Introducing the Scenario, a fundamental key concept in Taipy.
A Taipy Scenario represents a business problem with consistent data and parameters.
Scenarios are a powerful tool to create different versions of a business problem under different assumptions. This is especially valuable for what-if analysis in decision-making processes, enabling users to create, store, edit, and execute multiple scenarios with various parameters within the same application.
A scenario contains a Directed Acyclic Graph (or DAG) submittable for execution. The scenario DAG is a set of
tasks connecting data nodes together. It can also be broken down into smaller graphs for execution by defining
a Sequences. A sequence is a subset of connected tasks derived from the scenario's set of tasks, forming
a smaller executable DAG that can be submitted separately from the scenario DAG. A scenario can also contain
a set of additional data nodes that are not part of the scenario DAG to represent additional data related to
the scenario. The execution of the scenario does not compute the additional data nodes.
After analyzing its first scenario, an end-user may be interested in modifying input data nodes (not the intermediate nor the output data nodes), re-running the identical sequences or scenario and comparing the results with the previous run.
Once an initial scenario has been analyzed, users can modify the input data nodes (excluding the intermediates and output data nodes), rerun part of its tasks, and compare results.
This involves instantiating a second scenario, changing the input data, executing it, and comparing the outcomes with the first scenario.
This iterative process can be repeated across multiple scenarios, allowing for comprehensive exploration and analysis of different problem variations.
In the example
Here, our scenario consists of a graph of data nodes and tasks. The scenario graph can also be split into two sequences as subgraphs described earlier. The external light blue box in the flowchart below represents our scenario that contains all data nodes, tasks, and sequences.
Two use-cases arise from the utilization of Taipy scenarios:
- Use case 1 :¶
Each scenario represents a distinct instance of a business problem.
With Taipy, end-users can create, store, edit, and execute various scenarios within the same application.
Example
We build an application that:
-
first forecast the monthly demand for a plant
-
then, based on that forecast, generate the planning for the production orders.
The end-user creates the first scenario for January. It must contain everything the end-user needs to understand the January case, access input data, compute predictions, visualize our forecast algorithm results, make production decisions, and publish the January production orders.
Then the end-user creates another scenario for February using the new information provided for the February period. And so on.
- Use case 2:¶
Two scenarios represent the same business problem instance, but with different sets of assumptions.
Example
The end-user wants to simulate how the capacity data affects production planning for the February situation.
In the first scenario, we forecast demand and calculate production orders based on a low capacity assumption. In the second scenario, we assume a higher capacity value.
It's important to note that scenarios are not only useful for end-users, but also for data scientists. They can use scenarios as experiments to test different hypotheses. Essentially, scenarios provide a common concept that both data scientists and end-users can utilize.