# function.py
import numpy as np
import pandas as pd
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split, KFold

"""
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。
"""
id = "202210423209"
name = "万子玉"


# 这里写自己的代码实现
def sigmoid(z):
    return 1 / (1 + np.exp(-z))

def train_logreg(X, y, learning_rate=0.01, iterations=2000, lambda_reg=0.01, batch_size=32):
    X = np.insert(X, 0, 1, axis=1)  # 添加偏置项
    m, n = X.shape
    weights = np.random.randn(n) * 0.01  # 随机初始化权重

    for i in range(iterations):
        indices = np.arange(m)
        np.random.shuffle(indices)
        X_shuffled = X[indices]
        y_shuffled = y[indices]

        for j in range(0, m, batch_size):
            X_batch = X_shuffled[j:j + batch_size]
            y_batch = y_shuffled[j:j + batch_size]

            z = np.dot(X_batch, weights)
            predictions = sigmoid(z)
            gradient = np.dot(X_batch.T, (predictions - y_batch)) / X_batch.shape[0]

            # L2正则化
            gradient += (lambda_reg / m) * weights

            # 更新权重
            weights -= learning_rate * gradient

    return weights

    """
    对率回归训练函数。

    Args:
    - X (numpy array): 特征数据。
    - y (numpy array): 标签数据。
    - other_params (str): 其它参数，可选设置。若需设置其它参数，必须设置其默认值，否则将**无法通过**自动评测，**成绩为零**。

    Returns:
    - numpy array: 训练得到的权重。
    """


def predict_logreg(inX, weights):
    inX = np.insert(inX, 0, 1)  # 添加偏置项
    z = np.dot(inX, weights)
    prediction_prob = sigmoid(z)
    return 1 if prediction_prob >= 0.5 else 0


    """
    对率回归预测函数。

    Args:
    - inX (numpy array): 输入待分类的样本。
    - weights (numpy array): 训练得到的权重。

    Returns:
    - int: 预测的分类标签。
    """
def evaluate_model(X_val, y_val, weights):
    predictions = np.array([predict_logreg(x, weights) for x in X_val])
    accuracy = np.mean(predictions == y_val)
    return accuracy

def split_train_val(X, y, val_size=0.2):
    shuffled_indices = list(range(X.shape[0]))
    random.shuffle(shuffled_indices)
    X_shuffled = X[shuffled_indices]
    y_shuffled = y[shuffled_indices]
    
    val_count = int(X.shape[0] * val_size)
    
    X_val = X_shuffled[:val_count]
    y_val = y_shuffled[:val_count]
    X_train = X_shuffled[val_count:]
    y_train = y_shuffled[val_count:]
    
    return X_train, X_val, y_train, y_val



# 这里也写自己的代码实现
def load_data(filepath):
    with open(filepath, 'r') as file:
        lines = file.readlines()

    data = [line.strip().split() for line in lines if line.strip().split()[-1].isdigit()]
    data = [list(map(lambda x: float(x) if x != '?' else np.nan, row)) for row in data]

    # 移除类别标签缺失的样本
    data = [row for row in data if not np.isnan(row[-1])]

    # 使用“0”填充缺失值
    for row in data:
        for i in range(len(row)):
            if np.isnan(row[i]):
                row[i] = 0.0

    # 合并特征23的值2和3为0
    for row in data:
        if row[22] == 2 or row[22] == 3:  # 假设特征23是第23个元素
            row[22] = 0

    # 移除不需要的特征
    data = [row[:2] + row[3:23] + row[28:] for row in data]

    # 分离特征和标签
    X = [row[:-1] for row in data]
    y = [row[-1] for row in data]

    scaler = StandardScaler()
    X = scaler.fit_transform(X)

    return np.array(X), np.array(y)
    """
    从文件中加载数据，若不自行正确实现，则无法通过评测，**不记录成绩**。

    Args:
    - directory (str): 数据文件路径。

    Returns:
    - np.array: 特征数据。
    - np.array: 标签数据。
    """

if __name__ == "__main__":
    X_train, y_train = load_data("./data/horse-colic.data")
    X_test, y_test = load_data("./data/horse-colic.test")
    
    weights = train_logreg(X_train, y_train, learning_rate=0.01, iterations=2000)
    
    y_pred = predict_logreg(X_test[11], weights)
    print(f"预测结果: {y_pred}, 实际结果: {y_test[11]}")