def save_multiclass_classification(additional_metrics, model_desc,
model_path, fold_cnt, repeat_cnt):
max_metrics = additional_metrics["max_metrics"].transpose()
confusion_matrix = additional_metrics["confusion_matrix"]
with open(os.path.join(model_path, "README.md"), "w",
encoding="utf-8") as fout:
fout.write(model_desc)
fout.write("\n### Metric details\n{}\n\n".format(
max_metrics.to_markdown()))
fout.write("\n## Confusion matrix\n{}".format(
confusion_matrix.to_markdown()))
AdditionalMetrics.add_learning_curves(fout)
AdditionalMetrics.add_tree_viz(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_linear_coefs(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_permutation_importance(
fout, model_path, fold_cnt, repeat_cnt)
plots = additional_metrics.get("additional_plots")
if plots is not None:
AdditionalPlots.append(fout, model_path, plots)
AdditionalMetrics.add_shap_importance(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_shap_multiclass(fout, model_path, fold_cnt,
repeat_cnt)
fout.write("\n\n[<< Go back](../README.md)\n")
def save_binary_classification(additional_metrics, model_desc, model_path):
max_metrics = additional_metrics["max_metrics"].transpose()
confusion_matrix = additional_metrics["confusion_matrix"]
threshold = additional_metrics["threshold"]
with open(os.path.join(model_path, "README.md"), "w") as fout:
fout.write(model_desc)
fout.write("\n## Metric details\n{}\n\n".format(
max_metrics.to_markdown()))
fout.write("\n## Confusion matrix (at threshold={})\n{}".format(
np.round(threshold, 6), confusion_matrix.to_markdown()))
AdditionalMetrics.add_learning_curves(fout)
def save_multiclass_classification(additional_metrics, model_desc,
model_path):
max_metrics = additional_metrics["max_metrics"].transpose()
confusion_matrix = additional_metrics["confusion_matrix"]
with open(os.path.join(model_path, "README.md"), "w") as fout:
fout.write(model_desc)
fout.write("\n### Metric details\n{}\n\n".format(
max_metrics.to_markdown()))
fout.write("\n## Confusion matrix\n{}".format(
confusion_matrix.to_markdown()))
AdditionalMetrics.add_learning_curves(fout)
def save_multiclass_classification(additional_metrics, model_desc,
model_path, fold_cnt, repeat_cnt):
max_metrics = additional_metrics["max_metrics"].transpose()
confusion_matrix = additional_metrics["confusion_matrix"]
with open(os.path.join(model_path, "README.md"), "w") as fout:
fout.write(model_desc)
fout.write("\n### Metric details\n{}\n\n".format(
max_metrics.to_markdown()))
fout.write("\n## Confusion matrix\n{}".format(
confusion_matrix.to_markdown()))
AdditionalMetrics.add_learning_curves(fout)
AdditionalMetrics.add_tree_viz(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_linear_coefs(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_permutation_importance(
fout, model_path, fold_cnt, repeat_cnt)
AdditionalMetrics.add_shap_importance(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_shap_multiclass(fout, model_path, fold_cnt,
repeat_cnt)
def save_binary_classification(additional_metrics, model_desc, model_path,
fold_cnt, repeat_cnt):
max_metrics = additional_metrics["max_metrics"].transpose()
confusion_matrix = additional_metrics["confusion_matrix"]
threshold = additional_metrics["threshold"]
with open(os.path.join(model_path, "README.md"), "w") as fout:
fout.write(model_desc)
fout.write("\n## Metric details\n{}\n\n".format(
max_metrics.to_markdown()))
fout.write("\n## Confusion matrix (at threshold={})\n{}".format(
np.round(threshold, 6), confusion_matrix.to_markdown()))
AdditionalMetrics.add_learning_curves(fout)
AdditionalMetrics.add_tree_viz(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_linear_coefs(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_permutation_importance(
fout, model_path, fold_cnt, repeat_cnt)
AdditionalMetrics.add_shap_importance(fout, model_path, fold_cnt,
repeat_cnt)
AdditionalMetrics.add_shap_binary(fout, model_path, fold_cnt,
repeat_cnt)
fout.write("\n\n[<< Go back](../README.md)\n")
