Closed liu-jc closed 2 months ago
@liu-jc 1. can we reuse the evaluation pipeline of eval_util.evaluation.evaluate_forecasts and the evaluation metrics from gluonts.ev.metrics.x? it would be more consistent.
My idea is once you get the list of timesFM predictions, 1. you can covert it into a iterable object which batcher function support, 2. convert the forecast value into a SampleForecast object. https://ts.gluon.ai/stable/_modules/gluonts/model/forecast.html#SampleForecast
batcher
SampleForecast
@liu-jc 1. can we reuse the evaluation pipeline of eval_util.evaluation.evaluate_forecasts and the evaluation metrics from gluonts.ev.metrics.x? it would be more consistent.
My idea is once you get the list of timesFM predictions, 1. you can covert it into a iterable object which
batcherfunction support, 2. convert the forecast value into aSampleForecastobject. https://ts.gluon.ai/stable/_modules/gluonts/model/forecast.html#SampleForecast