If array-like, values used as weights to average the errors. Seasonal periodicity of the data multioutput or array-like of shape (n_outputs,), default=’uniform_average’ĭefines how to aggregate metric for multivariate (multioutput) data. Works with multioutput (multivariate) timeseries data In this case the function returns a large value ![]() Will not give infinite or undefined values unless the training data This metric is well suited to intermittent-demand series because it Used to compare forecast methods on a single series and also to compare ![]() Like other scaled performance metrics, this scale-free error metric can be MASE output is non-negative floating point. MeanAbsoluteScaledError # class MeanAbsoluteScaledError ( multioutput = 'uniform_average', multilevel = 'uniform_average', sp = 1 ) # MultioutputTimeSeriesRegressionForecaster
0 Comments
Leave a Reply. |