| name | data-aggregator |
| description | Aggregate tabular data by one or more dimensions with configurable metrics — sum, mean, count, min, max, median. Supports multi-level grouping and pivot table generation. |
| version | 0.8.0 |
Data Aggregator
Skill Summary
Performs group-by aggregations on tabular datasets. Supports single and multi-level grouping, configurable aggregate functions (sum, mean, count, min, max, median, std), and pivot table generation for cross-dimensional analysis. Can output flat aggregation tables or pivot matrices.
Aggregation Modes
- flat: Standard group-by → one row per group
- pivot: Cross-tabulation with row/column dimensions and value metric
- rolling: Time-based rolling window aggregation
Inputs
--input / -i: Path to input file (CSV or Parquet) (required)--group-by / -g: Grouping columns (comma-separated) (required)--metrics: Aggregation specs like revenue:sum,quantity:mean,order_id:count--mode: flat, pivot, or rolling (default: flat)--pivot-columns: Column for pivot table columns (for pivot mode)--pivot-values: Value column for pivot table (for pivot mode)--window: Window size for rolling mode--output / -o: Output path--sort-by: Sort result by specified column (descending)
Processing Steps
- Load dataset
- Parse metric specifications (column:function pairs)
- Apply group-by aggregation or pivot
- Sort results by specified column if provided
- Calculate group-level subtotals if multi-level grouping
- Save aggregated result
Output
- Aggregated dataset (CSV or Parquet)
- Console: Group count, top 10 groups by first metric
Implementation
- Command:
python3 ./skills/data_aggregator/aggregate.py -i <file> -g <columns> --metrics <specs>
- Dependencies:
pandas, numpy
