Method, Apparatus, and Computing Device for Constructing Prediction Model, and Storage Medium

A method comprising:

• obtaining, based on scenario information of a target prediction scenario or based on a target dataset of the target prediction scenario and the scenario information of the target prediction scenario, model search space corresponding to the target prediction scenario for which prediction is to be performed, wherein the model search space comprises models and hyperparameters;
• performing model training based on the target dataset and the models and the hyperparameters in the model search space, to obtain trained prediction models;
• and
• obtaining, based on evaluation results of the trained prediction models, a prediction model corresponding to the target prediction scenario from the trained prediction models;
• and
• wherein the scenario information of the target prediction scenario comprises a computing performance requirement and/or a prediction requirement, the computing performance requirement comprises one or more of memory information, central processing unit (CPU) information, or an inference speed, and the prediction requirement comprises prediction duration and/or a prediction periodicity for performing prediction for the target prediction scenario.

A method comprising:

• receiving a model obtaining request sent by a construction device, the model obtaining request requesting to obtain initial model search space corresponding to a target prediction scenario in which prediction is to be performed, and the initial model search space comprising models for prediction;
• determining, based on a historical dataset upon receiving the model obtaining request, models and hyperparameters that correspond to the target prediction scenario, wherein the historical dataset is a target dataset of the target prediction scenario, or data in the historical dataset is similar to data in the target dataset;
• and
• sending the initial model search space to the construction device, wherein the initial model search space comprises the models and the hyperparameters that correspond to the target prediction scenario.

A device comprising:

• at least one processor;
• and
• a non-transitory memory, coupled to the at least one processor and configured to store instructions that, when executed by the at least one processor, cause the device to:
• obtain, based on scenario information of a target prediction scenario or based on a target dataset of the target prediction scenario and the scenario information of the target prediction scenario, model search space corresponding to the target prediction scenario in which prediction is to be performed, wherein the model search space comprises models and hyperparameters;
• perform model training based on the target dataset and the models and the hyperparameters that are comprised in the model search space, to obtain trained prediction models;
• and
• obtain, based on evaluation results of the trained prediction models, a prediction model corresponding to the target prediction scenario from the trained prediction models;
• and
• wherein the scenario information of the target prediction scenario comprises a computing performance requirement and/or a prediction requirement, the computing performance requirement comprises one or more of memory information, central processing unit (CPU) information, or an inference speed, and the prediction requirement comprises prediction duration and/or a prediction periodicity for performing prediction for the target prediction scenario.