Draft: SaQC Machine Learning module

Module trying to make machine learning on timeseries do able in two liners.

Implements saqc.funcs.ml module, containing:

• ml.trainModel: Fit a model to explain target data by field data and store the model together with its configuration to a model folder

• ml.modelPredict: use stored model to predict field (processing function)

• ml.modelFlag: use stored (binary classification) model to flag field. (flagging dunction)

• ml.modelImpute: use stored model to apply imputation to field

Currently the Pipeline fails, because of a deprecation warning caused by a package (SHAP), wich is used by the Auto tuner Module (mljar-upervised).

saqc/core/modules/__init__.py

@@ -21,6 +21,7 @@ from saqc.core.modules.residuals import Residuals
 from saqc.core.modules.rolling import Rolling
 from saqc.core.modules.scores import Scores
 from saqc.core.modules.tools import Tools
+from saqc.core.modules.ml import Ml
 from saqc.core.modules.transformation import Transformation
 from saqc.core.register import FUNC_MAP
 
@@ -43,6 +44,7 @@ class FunctionsMixin(
     Scores,
     Tools,
     Transformation,
+    Ml,
 ):
     def _defer(self, fname, flocals):
         flocals.pop("self", None)