Quick Start =========== The ``functional_time_series`` package provides the ``functional_time_series.fts_spatial_entrypoint`` command-line entrypoint. It reads a GeoZarr dataset containing a time series, fits the spatial functional time series model, and writes the resulting curve coefficients as a GeoZarr dataset. Requirements ------------ Use Python 3.12 or newer, and install the project dependencies with Poetry: .. code-block:: bash poetry install Run the command from the repository root. For a list of arguments and their defaults: .. code-block:: bash poetry run python -m functional_time_series.fts_spatial_entrypoint --help Run the spatial FTS model ------------------------- The required arguments are the input data path and output path. The input should be a GeoZarr time series with a ``spatial_ref`` coordinate containing a valid ``crs_wkt`` attribute. Spatial data may be gridded (``x`` and ``y`` dimensions) or represented by a ``point_id`` dimension. For example, the repository includes a GRF output GeoZarr that can be used as input: .. code-block:: bash mkdir -p output poetry run python -m functional_time_series.fts_spatial_entrypoint \ --data_path tests/test_data/test_inputs/watson_runoff_racmo55_grf_output.zarr \ --output_path output/watson_runoff_racmo55_fts.zarr The input path may be a local path or a URL for a publicly available Zarr store. The output is written to a local GeoZarr store. The FTS output contains one variable for each analyzed input variable, with dimensions ``curve_id`` and ``coef_id``, plus the spatial reference. Options ------- The following options are optional: ``--curve_time_points`` Number of time steps in each curve (default: ``5``). Any trailing time steps that do not make a whole curve are omitted. ``--num_pca`` Number of principal components to retain in the first dimension-reduction step (default: ``2``). Only the first component of each point feeds the second step, so this does not change the result. ``--num_factors`` Number of factors, and therefore output coefficients, to retain in the second dimension-reduction step (default: ``3``). For example, to use 8 time steps per curve and retain 6 factors: .. code-block:: bash poetry run python -m functional_time_series.fts_spatial_entrypoint \ --data_path tests/test_data/test_inputs/watson_runoff_racmo55_grf_output.zarr \ --output_path output/watson_runoff_racmo55_fts_custom.zarr \ --curve_time_points 8 \ --num_factors 6