Skip to content

Public API

Every function below has its own reference and a runnable synthetic-data exploration. Start with a question, follow a story, or go straight to an argument. All pages describe Fieldwork 0.3.2 and its recorded source.

Use import fieldwork as fw. Analyses take pandas dataframes and preserve source values. Each walkthrough explains its population, shows real output, and includes a downloadable script that follows the same steps.

Your question Tool Walkthrough
What should I investigate first? fw.explore Municipal service requests and missing photos
Which measurements are available together? fw.missingness Marsh monitoring visits and station coverage
What determines this value, and what breaks the rule? fw.discover_dependencies Warehouse receipts and packaging conventions
What order makes this table readable? fw.suggest_paths Museum storage from collection to object
What conventions hide in populated values? fw.value_patterns Sensor exports, formats, and measurement relationships
Which values occur, and what does a short list omit? fw.levels Repair-cafe intake, outcomes, and omitted categories
What sits inside each group? fw.census Library borrowing and selected cohorts
What does a proposed key determine? fw.grain Books, editions, copies, and a binding discrepancy
How do two dimensions relate? fw.pairs Transit routes, tickets, and three kinds of absence
What counts sit behind one association? fw.joint_counts Bakery orders by pickup window and bread
How do my chosen dimensions fit together? fw.profile Service requests by channel, service, and photo
Which column roles deserve review? fw.infer_schema Manufactured parts and accidentally unique measurements
How has availability changed? fw.compare Festival registrations and a changing channel mix
Do two tables connect through a key? fw.relate Tool-library loans, members, and buddy pairs
Output Function Use it for
Bounded text fw.render_plaintext Terminal, notebook, and text reports.
Vector figure fw.render_svg Static documents and webpages.
Standalone report fw.render_html Sharing expandable evidence without Python.
Presentation dictionary fw.visualization_data A custom view of saved evidence.

Save result.to_dict() when you need an analytical archive. Presentation output can omit quantities or findings; it is not a substitute for the full result. See results and configuration for source-bound scopes, inspection, selection, recipes, and serialization.

These entry points link to complete argument tables, return contracts, and worked examples. explore and profile also document their option mappings explicitly.

Orient yourself in an unfamiliar table.

Understand which information arrives together.

Find mappings and the rows that break them.

Choose an order that makes a table readable.

Read the contents behind populated cells.

See the distribution of each field on its own.

Walk from broad groups to specific combinations.

Test what a supplied key really determines.

Separate observed mappings, association, and absence.

Inspect one two-way table without hiding its cells.

Explain a table through dimensions you have chosen.

Review column-role proposals before choosing dimensions.

Explain changes in availability between deliveries.

Check how two tables connect through a key.

Format a readable terminal report.

Create a static figure from saved evidence.

Share an explorable evidence report.

Use the labels and values behind a figure.

For runtime feedback and cancellation, see Progress, cancellation, and work budgets, including ProgressEvent, ProgressDisplay, CancellationToken, AnalysisCancelled, and the safe_errors control with its AnalysisError. Public signatures and editor documentation describe the supported controls; fieldwork.typing provides reusable option dictionary types.