stats-transformer

stats-transformer is a Python library for macroeconomic data transformation, analysis, and visualization. Built around a configuration-driven architecture, it handles data ingestion, resampling, feature engineering, and econometric modeling for time-series and panel datasets.

Features

• Feature Engineering: Advanced data transformations, frequency alignment, and robust merging capabilities for disparate datasets.
• Econometric Modeling: Built-in support for standard OLS, Robust OLS, Panel Regression, IV regression, discrete choice, time-series models, and unsupervised learning models (PCA, KMeans).
• Visualization: Automated generation of Exploratory Data Analysis (EDA) and regression model visual summaries (e.g., coefficient plots, residual plots, time-series tracking). Now includes a modular suite of standalone chart components for custom research plots.
• Configuration-Driven Orchestration: Fully integrated with YAML configuration (params.yaml) to enable reproducible, stage-based execution compatible with DVC pipelines.

Quickstart

1. Installation

To use it in your project via PyPI:

pip install stats-transformer

Or add it with uv:

uv add stats-transformer

For local development from this repository:

uv sync

2. Configuration (params.yaml)

Define your data sources, pipeline parameters, and model specifications in a params.yaml file:

data:
  featurization:
    entity_column: country
  datasets:
    - name: macro_data
      path: data/raw/macro_indicators.csv
      frequency: Q

model:
  model_type: panel_ols
  target_variable: gdp_growth
  independent_variables:
    - interest_rate
    - inflation

visualization:
  output_dir: reports/visualizations

3. Usage

Load a packaged example dataset:

from stats_transformer.data import list_examples, load_example

print(list_examples())
df = load_example("macrodb_gdp_inflation")

You can execute the pipeline via the command line using the Pipeline orchestrator:

# Run the full end-to-end pipeline
uv run python -m stats_transformer.pipeline --config params.yaml

Or you can interact with the API programmatically:

from stats_transformer import Pipeline

# Initialize the pipeline with your configuration
pipeline = Pipeline(params_path="params.yaml")

# Run specific stages sequentially
merged_data = pipeline.run(stage="resample")
transformed_data = pipeline.run(stage="features")
model_results = pipeline.run(stage="regression")

# Generate and save visualizations
pipeline.run(stage="visualization")

4. Testing

Verify the installation and library integrity by running the test suite:

uv run pytest tests

For more details on test coverage, see the Testing Suite.

Documentation

• Examples: For examples of running the models, see docs/validation/academic_examples.md.
• Visualization Walkthrough: For a guide on using the modular chart components, see notebooks/07_chart_components.ipynb.
• System Design: For more details on the system design, see docs/library/architecture.md.
• File Structure: For the standardized research folder structure, see docs/library/file_structure.md.
• Validation & Testing: For details on the testing suite, see docs/validation/testing_suite.md.

Agent Skill

Source checkouts of this repository include an optional stats-transformer-architecture agent skill for AI coding tools. It gives agents a compact map of the library architecture, pipeline stages, model contracts, and feature-engineering vocabulary.

The canonical skill source lives at .agents/skills/stats-transformer-architecture/. From the repository root, use the local terminal to run scripts/install-agent-skill.sh, which installs or refreshes tool-specific copies for Claude Code, OpenAI Codex, or Kilo Code:

./scripts/install-agent-skill.sh all

You can also target one tool at a time:

./scripts/install-agent-skill.sh claude
./scripts/install-agent-skill.sh codex
./scripts/install-agent-skill.sh kilo