dingxingdi Skill: multi-step-data-transformation-and-analytics-orchestration
1. Capability Definition & Real Case
- Professional Definition: The ability to complete enterprise data-workflow tasks that require multiple SQL and/or Python steps, intermediate result inspection, project-codebase navigation, and final artifact production rather than a single isolated query.
- Dimension Hierarchy: Data and ML Workflow Engineering->Enterprise Data Workflow Coding->multi-step-data-transformation-and-analytics-orchestration
Real Case
[Case 1]
- Initial Environment: A project workspace contains DBT-style project files, warehouse interfaces, documentation, and macros. The business request asks for a daily report on activity, leads, and opportunity status with intermediate model dependencies.
- Real Question: Implement the required project-level transformation so the daily sales activity output can be generated correctly from the existing project context.
- Real Trajectory: Inspect model files and macros, identify upstream sources and intermediate transformations, modify or add the necessary SQL models, run the project or targeted query steps, inspect intermediate outputs, and finalize the transformation pipeline.
- Real Answer: The project now produces the requested daily report through a correct multi-step transformation workflow grounded in the existing project structure.
- Why this demonstrates the capability: This capability is broader than single-query authoring because the agent must navigate a project, sequence multiple transformations, and reason over intermediate artifacts. The final answer emerges from workflow orchestration, not one standalone SQL statement.
[Case 2]
- Initial Environment: A bike-share analytics workspace includes a query interface, predefined answer-format files, and documentation for combining trip, station, and regional metadata. The task asks for enriched top-trip reporting under temporal constraints.
- Real Question: Construct the workflow needed to derive the requested enriched trip report, including all intermediate joins and derived fields.
- Real Trajectory: Probe the raw trip data, inspect auxiliary station and region sources, build intermediate SQL queries, validate partial outputs against the expected shape, and assemble the final result file with the required columns.
- Real Answer: A multi-step analytics workflow generates the final enriched trip report with correct joins, derived features, and output structure.
- Why this demonstrates the capability: The skill tested here is orchestration across multiple dependent steps. The agent must manage intermediate state and cumulative transformations, which is qualitatively harder than answering a one-query text-to-SQL prompt.
Pipeline Execution Instructions
To synthesize data for this capability, you must strictly follow a 3-phase pipeline. Do not hallucinate steps. Read the corresponding reference file for each phase sequentially:
Phase 1: Environment Exploration Read the exploration guidelines to discover raw knowledge seeds:
references/EXPLORATION.mdPhase 2: Trajectory Selection Once Phase 1 is complete, read the selection criteria to evaluate the trajectory:
references/SELECTION.mdPhase 3: Data Synthesis Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data:
references/SYNTHESIS.md