Coverage for addmo/s3_model_tuning/config/model_tuning_config.py: 100%

1import os 

2from pydantic import BaseModel, Field, PrivateAttr 

3from typing import Optional 

4from addmo.util.load_save_utils import root_dir 

5 

6 

7class ModelTunerConfig(BaseModel): 

8 

9 

10 models: list[str] = Field( 

11 ["ScikitMLP_TargetTransformed"], description="List of models to use (available models: ScikitMLP, ScikitLinearReg, ScikitLinearRegNoScaler, ScikitSVR, ScikitMLP_TargetTransformed, SciKerasSequential)", 

12 ) 

13 

14 trainings_per_model: int = Field( 

15 1, 

16 description="Number of trainings per model to choose the best " 

17 "from, this is done to avoid local minima" 

18 "during training.", 

19 ) 

20 

21 hyperparameter_tuning_type: str = Field( 

22 "OptunaTuner", 

23 description="Type of hyperparameter tuning, e.g., OptunaTuner, GridSearchTuner", 

24 ) 

25 hyperparameter_tuning_kwargs: Optional[dict[str, int]] = Field( 

26 default_factory=lambda: {"n_trials": 2}, 

27 description="Kwargs for the tuner. Set default value to n_trials: 2" 

28 ) 

29 

30 

31 validation_score_mechanism: str = Field( 

32 "cv", description="Validation score mechanism, e.g., cross validation, holdout" 

33 ) 

34 validation_score_mechanism_kwargs: Optional[dict[str, str]] = Field( 

35 default=None, description="Kwargs for the validation score mechanism" 

36 ) 

37 

38 validation_score_splitting: str = Field( 

39 "KFold", description="Validation score splitting, e.g., KFold, PredefinedSplit" 

40 ) 

41 validation_score_splitting_kwargs: Optional[dict[str, str]] = Field( 

42 default=None, description="Kwargs for the validation score splitter" 

43 ) 

44 

45 validation_score_metric: str = Field( 

46 "neg_root_mean_squared_error", 

47 description="Validation score metric, e.g., r2, neg_mean_absolute_error", 

48 ) 

49 validation_score_metric_kwargs: Optional[dict[str, str]] = Field( 

50 default=None, description="Kwargs for the validation score metric" 

51 ) 

52 

53 

54class ModelTuningExperimentConfig(BaseModel): 

55 def __init__(self, **data): 

56 super().__init__(**data) 

57 self._config_model_tuner = ModelTunerConfig() 

58 

59 name_of_raw_data: str = Field( 

60 "test_raw_data", description="Refer to the raw system_data connected to this" 

61 ) 

62 name_of_data_tuning_experiment: str = Field( 

63 "test_data_tuning", 

64 description="Refer to the system_data tuning experiment aka the input system_data for this model tuning experiment", 

65 ) 

66 name_of_model_tuning_experiment: str = Field( 

67 "test_model_tuning", description="Set name of the current experiment" 

68 ) 

69 abs_path_to_data: str = Field( 

70 os.path.join(root_dir(),'addmo_examples','raw_input_data','InputData.xlsx'), 

71 description="Path to the file that has the system_data", 

72 ) 

73 name_of_target: str = Field( 

74 "FreshAir Temperature", description="Name of the target variable" 

75 ) 

76 

77 # Model Tuning Variables 

78 start_train_val: str = Field( 

79 "2016-08-01 00:00", 

80 description="Start date and time for training and validation", 

81 ) 

82 stop_train_val: str = Field( 

83 "2016-08-14 23:45", description="Stop date and time for training and validation" 

84 ) 

85 start_test: str = Field( 

86 "2016-08-15 00:00", description="Start date and time for testing" 

87 ) 

88 end_test: str = Field( 

89 "2016-08-16 23:45", description="End date and time for testing" 

90 ) 

91 _config_model_tuner: Optional[ModelTunerConfig] = PrivateAttr() 

92