# function.py
import numpy as np

"""
实验说明:
- 自行编写实现数据加载与预测病马死亡的对率回归算法。

注意事项:
- 在 function.py 中实现对率回归算法。
- 可以自行导入其他库，需要在 requirements.txt 添加依赖，若未在 requirements.txt 中添加使用到的依赖，则无法自动评测，成绩为零。
- 自动评测脚本仅会加载 train_logreg、predict_logreg 和 load_data 函数，请勿修改函数名。

测评环境:
- CPU: Intel Xeon(R) CPU E3-1231 v3 @ 3.40GHz
- OS: Ubuntu 22.04.4 LTS
- Python: 3.10.12
- numpy: 2.1.1
"""


"""
请在这里填写学号和姓名，不填写即成绩无效。
姓名请直接**填写中文**
例如: name = "张三"
"""
id = "202210423629"
name = "tanjunfeng"


# 这里写自己的代码实现
import numpy as np
import pandas as pd


def load_data(train_path, test_path):
    """
    加载训练数据和测试数据

    Args:
        train_path (str): 训练数据文件路径
        test_path (str): 测试数据文件路径

    Returns:
        tuple: 包含训练数据特征、训练数据标签、测试数据特征、测试数据标签
    """
    train_data = pd.read_csv("./data/horse-colic.data", header=None, na_values='?')
    test_data = pd.read_csv("./data/horse-colic.test", header=None, na_values='?')

    # 处理缺失值，这里简单删除含有缺失值的行
    train_data = train_data.dropna()
    test_data = test_data.dropna()

    train_X = train_data.iloc[:, :-1].values
    train_y = train_data.iloc[:, -1].values
    test_X = test_data.iloc[:, :-1].values
    test_y = test_data.iloc[:, -1].values

    return train_X, train_y, test_X, test_y


def sigmoid(z):
    """
    Sigmoid函数

    Args:
        z (numpy.ndarray): 输入值

    Returns:
        numpy.ndarray: Sigmoid函数值
    """
    return 1 / (1 + np.exp(-z))


def train_logreg(train_X, train_y, learning_rate=0.01, num_iterations=1000):
    """
    训练对率回归模型

    Args:
        train_X (numpy.ndarray): 训练数据特征
        train_y (numpy.ndarray): 训练数据标签
        learning_rate (float, optional): 学习率. Defaults to 0.01.
        num_iterations (int, optional): 迭代次数. Defaults to 1000.

    Returns:
        numpy.ndarray: 训练好的模型权重
    """
    num_features = train_X.shape[1]
    num_samples = train_X.shape[0]

    # 初始化权重
    weights = np.zeros(num_features)

    for i in range(num_iterations):
        # 计算预测值
        z = np.dot(train_X, weights)
        h = sigmoid(z)

        # 计算梯度
        gradient = np.dot(train_X.T, (h - train_y)) / num_samples

        # 更新权重
        weights -= learning_rate * gradient

    return weights


def predict_logreg(test_X, weights):
    """
    使用训练好的模型进行预测

    Args:
        test_X (numpy.ndarray): 测试数据特征
        weights (numpy.ndarray): 训练好的模型权重

    Returns:
        numpy.ndarray: 预测结果
    """
    z = np.dot(test_X, weights)
    h = sigmoid(z)

    # 将概率值转换为类别预测0或1
    predictions = (h >= 0.5).astype(int)

    return predictions

if __name__ == "__main__":
    train_path = "./data/horse-colic.data"
    test_path = "./data/horse-colic.test"

    train_X, train_y, test_X, test_y = load_data(train_path, test_path)

    trained_weights = train_logreg(train_X, train_y)

    predictions = predict_logreg(test_X, trained_weights)

    # 可以在这里进一步评估预测结果的准确性等指标，例如：
    accuracy = np.mean(predictions == test_y)
    print("预测准确率:", accuracy)