Source code for bspump.model.model

import logging
import json

from bspump.asab import Configurable

###

L = logging.getLogger(__name__)

###




class Model(Configurable):
    """
    Generic `Model` object. Loads trained model and parameters.

    """

    ConfigDefaults = {
        "path_model": "",  # path to serialized model
        "path_parameters": "",  # path to serialized model
    }



    def __init__(self, app, id=None, config=None):
        self.Id = id if id is not None else self.__class__.__name__
        super().__init__("model:{}".format(self.Id), config=config)
        self.PathModel = self.Config["path_model"]
        self.PathParameters = self.Config["path_parameters"]
        self.App = app
        self.Loop = app.Loop




    def load_model_from_file(self):
        """
        Load model from file.
        """
        raise NotImplementedError()




    def load_parameters_from_file(self):
        """
        Loads model parameters from json file. Override if needed.
        """
        with open(self.PathParameters) as f:
            self.Parameters = json.load(f)




    async def update(self):
        """
        Updates model on fly.
        """
        pass




    def transform(self, *args):
        """
        Method used to transform data for model input.
        """
        raise NotImplementedError()




    def predict(self, *args):
        """
        Method uses model to predict value from sample.
        """
        raise NotImplementedError()