Module epiclass.utils.augment_predict_file

Augment a label prediction file with new metadata categories.

File header format important. Expects [md5sum, true class, predicted class, labels] lines. or [md5sum, true class, predicted class, split_nb, labels] lines.

Functions

def add_coherence(df: pd.DataFrame, category: str)

Add another metric based on multiple lines. Needs a file with EpiRR column.

https://stackoverflow.com/questions/17995024/how-to-assign-a-name-to-the-size-column https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.transform.html

def add_track_type_coherence(df)

Add a more complex coherence metric. Tells us how much tracks types agree on predicted class for a unique experiment

def augment_header(header, categories)

Augment the file header with new metadata categories

def augment_line(line, metadata: Metadata, categories: List[str], classes, split_nb_col: bool = False)

Augment a non-header line with new metadata labels and additional info on 2nd highest prob.

def augment_predict(metadata: Metadata, predict_path: Path, categories: List[str], append_name: str | None = None)

Read -> augment -> write, row by row.

Expects [md5sum, true class, predicted class, labels] lines.

Returns path of new file.

def correct_true(path: Path, category: str, metadata: Metadata)

Read file and replace 'True class' labels with metadata values for given category.

def main()

Augment a label prediction file with new metadata categories.

File header format important. Expects [md5sum, true class, predicted class, labels] lines.

def parse_arguments() ‑> argparse.Namespace

Return argument line parser.

def write_coherence(path, category: str)

Read file, add coherence for given category, write it updated to same path.