as we have added components to the PLANES scoring, we have included some strict stop() behavior. specifically, as written now the trend and shape components require at least 4x as many observations in the seed relative to the number of forecasted horizons evaluated.
related to #132 ... we are seeing that if some locations have fewer values then they may fail to run these components, while other locations in the data succeed.
we may consider a way to conditionally handle this failure so that some of the locations can be scored while others are "skipped" for that component.
noting that this may add significant complexity to downstream summarization, interpretation, and potential weighting for scores.
as we have added components to the PLANES scoring, we have included some strict
stop()behavior. specifically, as written now the trend and shape components require at least 4x as many observations in the seed relative to the number of forecasted horizons evaluated.related to #132 ... we are seeing that if some locations have fewer values then they may fail to run these components, while other locations in the data succeed.
we may consider a way to conditionally handle this failure so that some of the locations can be scored while others are "skipped" for that component.
noting that this may add significant complexity to downstream summarization, interpretation, and potential weighting for scores.