Referencing imported data in a model in a Decision Optimization experiment

Referencing imported data and defining solution output in a model

For Python DOcplex or OPL models, you can refer to imported data by using the following syntax.

Python models

For Python DOcplex models, the data that you imported in the Prepare data view in the experiment UI is accessible from the input dictionary. To access your imported data you must define each table with the following syntax inputs['tablename']. For example, in this line of code, food is an entity that is defined from the table called diet_food:

food = inputs['diet_food']

Similarly, to show tables in the Explore solution view of the experiment UI you must specify them using the syntax outputs['tablename']. For example,

outputs['solution'] = solution_df

defines an output table that is called solution. The entity solution_df in the Python model defines this table.

You can find this Diet example in the Model_Builder folder of the DO-samples. To import and run (solve) it in the experiment UI, see Solving and analyzing a Decision Optimization model: the diet problem.

OPL models

In an OPL model you must declare a tupleset, for each table that you imported in the Prepare data view and with the same names. The schema for each tupleset must have the same number of columns as the table and use the same field names. For example, if you have an input table in your Prepare data view called Product with the attributes name, demand, insideCost, and outsideCost, your OPL model must contain the following definition:

tuple TProduct {
   key string name;
   float demand;
   float insideCost;
   float outsideCost;
 };

{TProduct}     Product = ...;

The limitation on only using tuples and tuple sets as OPL input is to facilitate integration with data sources. For example, SQL data sources can be accessed and data-streamed with a minimum of effort; NoSQL data sources can be accessed and data can be transformed automatically to tables. If necessary, the optimization model developer can reformulate the data to populate other data structures during the optimization, but this manipulation must not affect the input or output data.

Similarly, if you want to display a table in the Explore solution view, you must define a tupleset for this output table in your OPL model. For example, this code produces an output table with 3 columns in the solution.

/// solution
 tuple TPlannedProduction {
   key string productId;
   float insideProduction;
   float outsideProduction;
 }

{TPlannedProduction} plan = {<p.name, Inside[p], Outside[p]> | p in Products};

You can find this example OPL model for a pasta production problem in the Model_Builder folder of the DO-samples. You can download and extract all the samples. Select the relevant product and version subfolder.

Learn more

For a tutorial on creating Python DOcplex models, see Solving and analyzing a Decision Optimization model: the diet problem.
For more information about OPL models, see Decision Optimization OPL models.