Data Analytics
Generate browser-based Python notebooks for audit data analytics — population testing, sampling, exception classification, and Excel reporting with full audit trail.
How It Works
Data Analytics generates interactive Python notebooks that run entirely in the browser. Describe your audit test—control objective, data sources, population logic, and exception criteria—and the AI produces a notebook config (JSON) that you import into the browser-based template. Drop your CSV files, click Run All, and download formatted Excel results with a full audit trail.
Key Features
- AI generates the complete analytic—preprocessing, population building, sampling, testing, and Excel export
- Browser-based Python notebooks with syntax highlighting (CodeMirror) and live output
- Multi-file support—upload multiple CSVs for cross-referencing and matching
- Reproducible random sampling with configurable seed
- Auto-generated Excel workbook: Summary, Population, Sample Test Results, and Audit Log
- Full audit trail—every cell's source code is captured in the Excel output
- Import/Save—share notebook configs as JSON or standalone HTML files
What You Need
Tell the AI what you want to test: the control objective, data sources (what columns to expect), how to build the population, and what determines pass vs. fail. The AI handles the rest.
One or more CSV exports from your systems—ERP extracts, transaction logs, configuration exports, access reports, or any tabular data.
Control narratives, RCMs, policy documents, or business rules that define expected behavior. The AI uses these to design more precise test logic.
Tutorial: Build a Data Analytic
Download the Data Analytics skill and add it to your Claude Code or Claude Desktop skills directory. The skill is available via slash command or triggers automatically when you describe an audit data analytic.
Tell Claude what you want to test. Include the control objective, what data you have, how to identify the population, and what success looks like. Claude will ask clarifying questions if needed, then generate the notebook JSON.
Open the notebook template, click Import JSON, and paste the generated config. The notebook rebuilds with your custom cells, file inputs, and test logic.
Upload your CSV file(s) into the drop zones, click Run All, and download the Excel results. Every cell is editable—adjust parameters, refine logic, and re-run as needed.
Supported Analytic Patterns
- Population Sampling — Filter, sample, and test transactions or events
- Multi-Source Matching — Join and compare data from multiple systems
- Completeness Testing — Verify all expected items exist
- Threshold / Outlier Detection — Flag items exceeding policy limits
- Rule-Based Compliance — Apply business rules and classify exceptions
- Duplicate Detection — Find records that should be unique
- Timeliness Testing — Measure elapsed time against SLA requirements