def from_dfs(cls, Xdf, ydf, ML=None, bounds=None, features=None):
X = Xdf.values
y = ydf.values
preprocessor = Rescaler.from_data(X)
postprocessor = Normalizer.from_data(y)
if not ML: from sklearn.svm import SVR as ML
if ydf.ndim == 1:
predictor = ML()
predictor.metric = ydf.name
else:
def create_predictor(metric):
ml = ML()
ml.metric = metric
return ml
predictors = [create_predictor(col) for col in ydf]
predictor = MultiPredictor(predictors)
if not features: features = tuple(Xdf)
self = cls(predictor, preprocessor, postprocessor, bounds, features)
self.fit(X, y)
return self
def create_predictor(metric):
ml = ML()
ml.metric = metric
return ml
