Benchmarking¶

Benchmarking is used to test the 3D CNN model with standard machine learning models. The utility comes with existing models but the user can add, remove or tweak models based on his requirement. Refer scikit learn for more information about the models: https://scikit-learn.org/stable/supervised_learning.html#supervised-learning

dlmfg.utilities.benchmarking.benchmarking_models(max_models)[source]¶

benchmarking_models returns a list of models and model names less that or equal to max_models

Parameters: max_models (int (required)) – maximum number of models
Returns: bn_models: list of models used for benchmarking
Return type: list
Returns: bn_models_name: list of model names used for benchmarking
Return type: list

dlmfg.utilities.benchmarking.benchmarking_models_eval(bn_models, point_data, kcc_dataset, assembly_kccs, bm_path, test_size)[source]¶

benchmarking_models_evals trains each of the model based on the dataset and returns

Parameters

bn_models – list of models to be benchmarked
point_data (numpy.array (samples*nodes) (required)) – input data consisting of node deviations
kcc_dataset (numpy.array (samples*kccs) (required)) – output data consisting of process parameters/KCCs
assembly_kccs (float (required)) – number of assembly KCCs
bm_path (str (required)) – Benchmarking path to save benchmarking results
test_size – The test size split

Returns

bn_metrics_eval: Benchmarking metrics

Return type

numpy.array [bn_models*kccs*metrics]