compute_accuracy.py

import os
import operator
from typing import List

def load_dataset(path) -> List[tuple[List[str], str]]:
    result = []
    for file in os.listdir(path):
        if file.endswith('.txt'):
            with open(path + '/' + file) as f:
                cluster_reading = True
                cluster = []
                outliers = []
                for line in f:
                    line = line.rstrip()
                    if len(line) > 0:
                        if not line.startswith('#'):
                            if cluster_reading:
                                cluster.append(line)
                            else:
                                outliers.append(line)
                    else:
                        cluster_reading = False
                for outlier in outliers:
                    result.append((cluster, outlier))
    return result


def compute_accuracy(sim_fun, log_failed_cases=True):
    dataset = load_dataset('evaluation/outlier-detection')
    similarities = {}
    correct_guesses = 0
    total_guesses = 0
    for sample in dataset:
        cluster, outlier = sample
        items = cluster + [outlier]
        compactness = {}
        for item1 in items:
            compactness[item1] = 0.0
            for item2 in items:
                if item1 != item2:
                    compactness[item1] += sim_fun(item1, item2)
        least_compact = sorted(compactness.items(), key=operator.itemgetter(1))[0][0]
        if least_compact == outlier:
            correct_guesses += 1
        elif log_failed_cases:
            print(f"Failed {cluster} / {outlier}")
        total_guesses += 1
    return correct_guesses / total_guesses