Module epiclass.utils.extract_metrics

Extract metrics for other classifiers output files, or csv output of cometML for neural networks.

Functions

def extract_neural_network_metrics(file_path: str) ‑> pandas.core.frame.DataFrame

Extracts validation metrics for a Neural Network classifier from a CSV file (comet ML export)

This function reads a CSV file, extracts the split number and validation metrics, and stores these metrics in a DataFrame. The CSV file is expected to contain columns for different validation metrics and a column 'Name' that includes information about the split number.

Args

file_path : str

The path to the CSV file.

output_file_path : str

The path of the output file.

Returns

pd.DataFrame

A DataFrame with one row for each metric of each split. The DataFrame contains the following columns: 'classifier', 'split', 'metric', and 'value'.

def extract_other_estimators_metrics(file_path: str) ‑> pandas.core.frame.DataFrame

Extracts validation metrics for multiple classifiers from a log file.

This function reads a log file, identifies lines that contain validation metrics, and stores these metrics in a DataFrame. The log file is expected to contain one or more sections for each classifier, each section containing lines for different splits and their associated validation metrics.

Args

file_path : str

The path to the log file.

Returns

pd.DataFrame

A DataFrame with one row for each metric of each split of each classifier. The DataFrame contains the following columns: 'classifier', 'split', 'metric', and 'value'.

def main()

Main function

def parse_arguments() ‑> argparse.Namespace

Add an argument parser