---

name: data-and-io description: "Guides Pyomo users through AbstractModel loading, DataPortal input, and file-backed model initialization." disable-model-invocation: true metadata: disco-role: operating license: NOASSERTION

Data and IO

Use this sub-skill when the task begins with a data file, a spreadsheet, a configuration file, or an AbstractModel that needs create_instance().

What this sub-skill covers

• AbstractModel and create_instance() workflows.
• DataPortal and file-backed model initialization.
• .dat, .tab, Excel, YAML, and JSON inputs.
• Tutorial-style data loading and model instantiation.

What to route elsewhere

• Pure model construction without external data goes to modeling-basics.
• pyomo solve, pyomo convert, and config-template questions go to solve-and-cli.
• GDP, DAE, network, MPEC, and units go to structured-modeling.
• Solver/back-end or GUI dependency issues go to solver-extensions.

Read these references

• ../../references/data-and-io.md for verified file-loading patterns and data families.
• ../../references/cli-reference.md when the same data file is passed through a Pyomo CLI command.
• ../../references/troubleshooting.md when a load fails because of a missing package, a bad namespace, or a schema mismatch.

Use this helper

• ../../scripts/check_import.py for a quick import and metadata sanity check before debugging a data-loading issue.

Typical workflow

1. Declare an AbstractModel with the sets and parameters the data will fill.
2. Map the data file structure to the model indices and domains.
3. Load the data with create_instance() or DataPortal.
4. Inspect the instantiated model with pprint().
5. Only then connect the instance to a solver or transformation.

Common request patterns

• "How do I load a .dat file into my model?"
• "How do I use namespaces with DataPortal?"
• "How do I read Excel or tabular inputs into Pyomo?"
• "Why does create_instance() complain about missing indices?"
• "How do I generate or read a JSON/YAML config for a Pyomo run?"

Common failure modes

• A set dimension in the data does not match the model declaration.
• A parameter value is missing or appears under the wrong namespace.
• Spreadsheet helpers are unavailable because optional packages are missing.
• A file path is wrong, relative to the wrong working directory, or only valid inside the original checkout.

Practical guidance

• Keep the initial data example as small as possible.
• Verify the schema before scaling to a large file.
• Prefer explicit namespaces and explicit indices when multiple data blocks are involved.
• Use the tutorial-style examples as a reference pattern, but do not depend on the original repository path at runtime.

Examples of what belongs here

• AbstractModel plus one .dat file.
• DataPortal loading from a table or spreadsheet.
• A CLI invocation that loads model data from a config file.

Examples of what does not belong here

• Solver selection and execution flags.
• DAE, GDP, network, or complementarity workflows.
• Optional solver or GUI backends that are unrelated to file loading.

Related routes

• Move to modeling-basics when the data file is not the main problem.
• Move to solve-and-cli when the task is about CLI execution after data load.
• Move to structured-modeling when the model family is GDP, DAE, network, MPEC, or units.