[Discussion] How to group models if there are multiple hyperparams for the same model?

[AbhinavGor]

Two options:

• Do we have the same model module_id, version n number of times for n number of hyperparam settings?

models:
-   id: 25
    version: 1
    hyperparameters:
        criterion: squared_error
        splitter: best
        max_depth: null
        min_samples_split: 2
        min_samples_leaf: 1
        min_weight_fraction_leaf: 0.0
        max_features: 1
        random_state: null
        max_leaf_nodes: null
        min_impurity_decrease: 0.0
        ccp_alpha: 0.0
-   id: 25
    version: 1
    hyperparameters:
        criterion: squared_error
        splitter: best
        max_depth: null
        min_samples_split: 2
        min_samples_leaf: 1
        min_weight_fraction_leaf: 0.0
        max_features: 3
        random_state: null
        max_leaf_nodes: null
        min_impurity_decrease: 0.0
        ccp_alpha: 0.0
-   id: 25
    version: 1
    hyperparameters:
        criterion: squared_error
        splitter: best
        max_depth: null
        min_samples_split: 2
        min_samples_leaf: 1
        min_weight_fraction_leaf: 0.0
        max_features: 5
        random_state: null
        max_leaf_nodes: null
        min_impurity_decrease: 0.0
        ccp_alpha: 0.0
-   id: 25
    version: 1
    hyperparameters:
        criterion: squared_error
        splitter: best
        max_depth: null
        min_samples_split: 2
        min_samples_leaf: 1
        min_weight_fraction_leaf: 0.0
        max_features: 7
        random_state: null
        max_leaf_nodes: null
        min_impurity_decrease: 0.0
        ccp_alpha: 0.0

• Do we define a model moduel_id, version once and put all the n hyperparam setttings that we want to test with this model?

models:
-   id: 25
    version: 1
    hyperparameters:
       - 1: 
            criterion: squared_error
            splitter: best
            max_depth: null
            min_samples_split: 2
            min_samples_leaf: 1
            min_weight_fraction_leaf: 0.0
            max_features: 1
            random_state: null
            max_leaf_nodes: null
            min_impurity_decrease: 0.0
            ccp_alpha: 0.0
       - 2:
            criterion: squared_error
            splitter: best
            max_depth: null
            min_samples_split: 2
            min_samples_leaf: 1
            min_weight_fraction_leaf: 0.0
            max_features: 3
            random_state: null
            max_leaf_nodes: null
            min_impurity_decrease: 0.0
            ccp_alpha: 0.0
       - 3:
            criterion: squared_error
            splitter: best
            max_depth: null
            min_samples_split: 2
            min_samples_leaf: 1
            min_weight_fraction_leaf: 0.0
            max_features: 5
            random_state: null
            max_leaf_nodes: null
            min_impurity_decrease: 0.0
            ccp_alpha: 0.0
       - 4:
            criterion: squared_error
            splitter: best
            max_depth: null
            min_samples_split: 2
            min_samples_leaf: 1
            min_weight_fraction_leaf: 0.0
            max_features: 7
            random_state: null
            max_leaf_nodes: null
            min_impurity_decrease: 0.0
            ccp_alpha: 0.0

[AbhinavGor]

Ongoing discussion with @prat-man @kapkic