model_script/model/ros_multi_class_model_predice_analyse.py

import numpy as np


def parse_line(line):
    """ 解析每一行数据 """
    parts = line.strip().split("\t")
    label = int(parts[0])
    scores = np.array([float(x) for x in parts[2].strip("[]").split(",")])

    # 找到最大值索引
    max_index = np.argmax(scores)

    # 生成 (label, score) 形式的 aucs
    aucs = np.array([(1 if i == label else 0, scores[i]) for i in range(len(scores))])

    # 生成 (是否为真实 label, 是否为最大值) 的 accuracyRate
    accuracy_rate = np.array([(1 if i == label else 0, 1 if i == max_index else 0) for i in range(len(scores))])

    return aucs, accuracy_rate


def compute_auc(auc_data):
    """ 计算 AUC """
    num_classes = len(auc_data[0])  # 8 classes
    auc_scores = []

    for i in range(num_classes):
        col_data = np.array([row[i] for row in auc_data])  # 取第 i 列
        labels, scores = col_data[:, 0], col_data[:, 1]

        # 按 scores 降序排序
        sorted_indices = np.argsort(-scores)
        sorted_labels = labels[sorted_indices]

        # 计算正负样本数
        pos = np.sum(sorted_labels == 1)
        neg = len(sorted_labels) - pos
        if pos == 0 or neg == 0:
            auc_scores.append(0.5)
            continue

        # 计算 AUC
        rank_sum = np.sum(np.where(sorted_labels == 1)[0] + 1)  # 计算正样本的秩次和
        auc = (rank_sum - pos * (pos + 1) / 2) / (pos * neg)
        auc_scores.append(auc)

    return auc_scores


def compute_accuracy_rate(acc_data):
    """ 计算 accuracy """
    num_classes = len(acc_data[0])  # 8 classes

    # 全局 accuracy 计算
    acc_flatten = np.vstack(acc_data)
    global_correct = np.sum((acc_flatten[:, 0] == 1) & (acc_flatten[:, 1] == 1))
    total_count = acc_flatten.shape[0]
    global_accuracy = global_correct / total_count

    # 按 label 计算 accuracy
    per_label_accuracy = []
    for i in range(num_classes):
        col_data = np.array([row[i] for row in acc_data])  # 取第 i 列
        correct = np.sum((col_data[:, 0] == 1) & (col_data[:, 1] == 1))
        total_positive = np.sum(col_data[:, 0] == 1)
        per_label_accuracy.append(0 if total_positive == 0 else correct / total_positive)

    return global_accuracy, per_label_accuracy


if __name__ == "__main__":
    file_path = "/Users/zhao/Desktop/tzld/ros/ros_predict_20250302.txt"  # 本地文件路径

    # 读取数据
    with open(file_path, "r") as f:
        data_lines = f.readlines()

    # 解析数据
    parsed_data = [parse_line(line) for line in data_lines]
    auc_data = [item[0] for item in parsed_data]
    acc_data = [item[1] for item in parsed_data]

    # 计算 AUC
    auc_scores = compute_auc(auc_data)

    # 计算 Accuracy
    global_acc, per_label_acc = compute_accuracy_rate(acc_data)

    # 打印结果
    print("AUC Scores:")
    for i, auc in enumerate(auc_scores):
        print(f"Label {i}: AUC = {auc:.4f}")

    print(f"\nGlobal Accuracy: {global_acc:.4f}")

    print("\nPer Label Accuracy:")
    for i, acc in enumerate(per_label_acc):
        print(f"Label {i}: Accuracy = {acc:.4f}")