mllam-verification

This package exposes various plotting routines for verification of prediction data relative to some reference data. The plotting functions are meant to be used in JupyterNotebooks or Python scripts, and thus no CLI is provided.

The plotting routines accept a handle to a callable, which is used for calculating the statistical metric to plot. Let's take the plot_single_metric_timeseries function as an example:

from mllam_verification.plot import plot_single_metric_timeseries
from mllam_verification.operations.statistics import rmse

ax = plot_single_metric_timeseries(
    da_reference,
    da_prediction,
    time_axis="UTC",
    stats_operation=rmse,
    hue="analysis_time",
)

The plot_single_metric_timeseries function will call the rmse function to compute the RMSE between the reference and prediction data before plotting. An axes object is returned, which can be used to add more plots to the same figure, or to adjust the axes layout.

The following table outlines the different types of plots one can produce with the mllam-verification package and what they support. More to come!

Name	Plot function name	Example	Grouped	Elapsed	UTC	Multi	Multi model	Multi variable	Point	Regular
Single metric timeseries	plot_single_metric_timeseries		✅¹	✅¹	✅	✅	❌	❌	✅	✅
Single metric hovmöller	plot_single_metric_hovmoller		✅	✅	✅	✅	❌	❌	✅	✅
Single metric gridded map	plot_single_metric_gridded_map		✅	❌	❌	❌	❌	❌	❌	✅
single metric point map²	plot_single_metric_point_map ²		✅	❌	❌	✅²	✅²	❌	✅	✅

¹ without persistence
² not supported yet, but soon to come
³ maybe not a good idea e.g. if points overlap in grid.

Developing mllam-data-prep

To work on developing mllam-verification it is easiest to install and manage the dependencies with uv. To get started clone your fork of the main repo locally:

git clone https://github.com/<your-github-username>/mllam-verification
cd mllam-verification

You can now e.g. run tests of mllam-verification with uv directly

uv run pytest

or you can first create a virtualenv and install the dependencies

uv venv
uv sync --all-extras
uv run pytest

All the linting is handelled by pre-commit which can be setup to automatically be run on each git commit by installing the git commit hook:

uv run pre-commit install

Then branch, commit, push and make a pull-request :)

Repo structure

.
├── mllam_verification
│   ├── operations
│   │   ├── __init__.py
│   │   ├── array_handling.py           # Contains helper functions for array operations
│   │   └── statistics.py               # Contains functions for computing statistics e.g. mean, std, etc.
│   ├── __init__.py
│   ├── __main__.py                     # Entry point of the package
│   ├── plot.py                         # Main script for producing plots
│   └── validation.py                   # Contains functionality to validate function arguments
├── uv.lock
├── pyproject.toml
└── README.md