MONET Stats 🍃⚡

A comprehensive statistics and utility library for atmospheric sciences, optimized for the Pangeo ecosystem and fully Aero Protocol compliant.

Core Mission

Architect scientific pipelines that balance four competing goals:

• Speed: Aggressive vectorization (Numpy/Xarray) and lazy evaluation (Dask).
• Maintainability: Strictly typed code with NumPy-style docstrings.
• Provenance: Automatically track data lineage (what happened to the data) via attrs['history'].
• Visualization: A hybrid approach (Matplotlib for papers, HvPlot for interaction).

Installation

pip install monet-stats

Optional dependencies for Pangeo stack:

pip install monet-stats[dask,docs,test]

Quick Start

Error Metrics

import xarray as xr
import numpy as np
from monet_stats.error_metrics import MB, RMSE

# Assume all data > RAM. Use dask chunks immediately.
obs = xr.open_dataset('obs.nc', chunks={'time': 100})['variable']
mod = xr.open_dataset('mod.nc', chunks={'time': 100})['variable']

# Compute Mean Bias map over the time dimension
bias = MB(obs, mod, axis='time')

# Automatic provenance tracking
print(bias.attrs['history'])

Categorical Skill Scores

from monet_stats.contingency_metrics import HSS, ETS

# Evaluate Heidke Skill Score at a specific threshold
skill = HSS(obs, mod, minval=50.0)

Documentation

Full API documentation and tutorials are available at: https://noaa-oar-arl.github.io/monet-stats

CI/CD and Code Quality

This project uses a comprehensive CI/CD pipeline with the following quality checks:

• Testing: Multi-Python version testing (3.8-3.12) with 60%+ coverage
• Code Formatting: Black and Ruff formatting enforcement
• Linting: Ruff and Pycodestyle linting
• Type Checking: MyPy static type analysis
• Provenance: Mandatory history attribute updates on all transformations