Goal: Statistics for differening >1 predictions in time (typically) that have the same coordinates.
[ ] Create a notebook that trains an estimator with an earlier phase of data and predicts with later portions of the data set using the model trained on early data, allowing mapping of shifts in land use / land cover or some other pixel-specific classification, regression, or clustering predictions over time.
[ ] With or without polygon filters for zonal statistics
[ ] For categorical predictions, creation transition matrices (fractional transition of each class / cluster to another between time steps)
Goal: Statistics for differening >1 predictions in time (typically) that have the same coordinates.
This work may take place in
xarray_filters