Another big update

[kaylode]

• Add Optuna for hyperparameters tuning, resolves Hyperparamter tuning #31
• Update YAML reader, allows recursive configs
• Update dependencies requirement
• Add DVC for storing sample data
• Add CML for auto report
• Add Docker workflows for Github Action
• Fix several known issues

[github-actions]

Tabular Classification Report

Metrics

acc	bl_acc	weighted-f1	weighted-precision	weighted-recall	iters
0.79688	0.76318	0.79182	0.79484	0.79688	0

SHAP train

SHAP val

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params_gamma	params_learning_rate	params_max_depth	params_n_estimators	params_reg_alpha	params_reg_lambda	best_key	model_name	state
0	0.7768013037	1680459169426	1680459170072	645	0.7647683603	0.0436961066	6	134	0.9552591038	0.2987225422	bl_acc	xgboost	COMPLETE
1	0.829472704	1680459170075	1680459170315	240	0.2727798881	0.6248019055	2	219	0.2329221446	0.6442941877	bl_acc	xgboost	COMPLETE
2	0.7930392271	1680459170317	1680459170554	237	0.8881550858	0.8783533478	3	175	0.3800789093	0.7175942691	bl_acc	xgboost	COMPLETE

Figures

[github-actions]

Classification Report

Metrics

acc	bl_acc	weighted-f1	iters
0.5	0.5	0.5	10

Confusion Matrix

Errorcases

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params	best_key	model_name	state
0	0.25	1680459152917	1680459163744	10826	0.000894301	bl_acc	efficientnet_b0	COMPLETE
1	0	1680459163746	1680459174483	10737	0.0003934934	bl_acc	efficientnet_b0	COMPLETE

Figures

[github-actions]

Semantic Report

Metrics

precision	recall	dice	iters
0.19583	1	0.32743	2
0.19583	1	0.32743	4
0.19681	1	0.32881	6
0.24579	1	0.39452	8
0.53645	0.99993	0.69825	10

Prediction

[github-actions]

Tabular Classification Report

Metrics

acc	bl_acc	weighted-f1	weighted-precision	weighted-recall	iters
0.79688	0.76318	0.79182	0.79484	0.79688	0

SHAP train

SHAP val

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params_gamma	params_learning_rate	params_max_depth	params_n_estimators	params_reg_alpha	params_reg_lambda	best_key	model_name	state
0	0.7813991386	1680462109244	1680462109508	264	0.5371239379	0.0154546728	7	335	0.9695597913	0.1647784531	bl_acc	xgboost	COMPLETE
1	0.7813991386	1680462109510	1680462109738	228	0.1418470782	0.5631946884	9	461	0.492812173	0.0596589358	bl_acc	xgboost	COMPLETE
2	0.8236526598	1680462109739	1680462109988	248	0.1051527341	0.4155308767	3	293	0.6729336491	0.4558153184	bl_acc	xgboost	COMPLETE

Figures

[github-actions]

Semantic Report

Metrics

precision	recall	dice	iters
0.19583	1	0.32743	2
0.19583	1	0.32743	4
0.19681	1	0.32881	6
0.24579	1	0.39452	8
0.53645	0.99993	0.69825	10

Prediction

[github-actions]

Classification Report

Metrics

acc	bl_acc	weighted-f1	iters
0.5	0.5	0.5	10

Confusion Matrix

Errorcases

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params_optimizer.args.lr	best_key	model_name	state
0	0	1680462134695	1680462151043	16348	0.0005177835	bl_acc	efficientnet_b0	COMPLETE
1	0.25	1680462151045	1680462166954	15908	0.0003340079	bl_acc	efficientnet_b0	COMPLETE

[github-actions]

Semantic Report

Metrics

precision	recall	dice	iters
0.19583	1	0.32743	2
0.19583	1	0.32743	4
0.19681	1	0.32881	6
0.24579	1	0.39452	8
0.53645	0.99993	0.69825	10

Prediction

[github-actions]

Tabular Classification Report

Metrics

acc	bl_acc	weighted-f1	weighted-precision	weighted-recall	iters
0.79688	0.76318	0.79182	0.79484	0.79688	0

SHAP train

SHAP val

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params_gamma	params_learning_rate	params_max_depth	params_n_estimators	params_reg_alpha	params_reg_lambda	best_key	model_name	state
0	0.7784891165	1680464786259	1680464786501	242	0.4074474413	0.2654992962	3	101	0.9838639368	0.9227559496	bl_acc	xgboost	COMPLETE
1	0.7673146316	1680464786504	1680464786805	301	0.6753901107	0.6567590861	9	392	0.2329300361	0.631131902	bl_acc	xgboost	COMPLETE
2	0.7602723781	1680464786809	1680464787056	246	0.0134278736	0.0766996811	2	454	0.6767866185	0.7167346956	bl_acc	xgboost	COMPLETE

Figures

[github-actions]

Classification Report

Metrics

acc	bl_acc	weighted-f1	iters
0.5	0.5	0.5	10

Confusion Matrix

Errorcases

Hyperparameters Tuning

Leaderboard

number	value	datetime_start	datetime_complete	duration	params_optimizer.args.lr	best_key	model_name	state
0	0.25	1680464764840	1680464778849	14008	0.0007045775	bl_acc	efficientnet_b0	COMPLETE
1	0.25	1680464778851	1680464792728	13877	0.0007697312	bl_acc	efficientnet_b0	COMPLETE