继续完成所有的代码，给出最终代码

为了帮助您完成一个基于多元线性回归的房价预测项目，并采用小批量随机梯度下降法进行训练，以下是结合给定指导的一个完整示例代码框架：

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import mean_squared_error, r2_score

# 假设 `load_data()` 函数可以从您的数据文件中加载数据。
def load_data(file_path):
    # 这里应替换为您实际加载数据的方法，例如用pandas读取CSV等
    pass

# 加载数据集
data = load_data('path_to_your_data.csv')

# 将数据集划分为训练集和测试集
train_set, test_set = train_test_split(data, test_size=0.2, random_state=42)

# 数据标准化处理
scaler = StandardScaler()
scaler.fit(train_set.drop("house_price", axis=1))  # 假设目标变量名为"house_price"
X_train_scaled = scaler.transform(train_set.drop("house_price", axis=1))
y_train = train_set["house_price"]
X_test_scaled = scaler.transform(test_set.drop("house_price", axis=1))
y_test = test_set["house_price"]

# 初始化模型参数 (这里以简单的方式直接设置初始权重和偏置)
w = np.zeros(X_train_scaled.shape[1])
b = 0

# 定义模型
def linreg(X, w, b):
    return np.dot(X, w) + b

# 定义损失函数 - 均方误差
def squared_loss(y_hat, y):
    return (y_hat - y.reshape(y_hat.shape)) ** 2 / 2

# 定义小批量随机梯度下降法
def sgd(params, lr, batch_X, batch_y):
    m = len(batch_y)
    grad = 2 * np.dot((linreg(batch_X, *params) - batch_y), batch_X) / m
    params[0] -= lr * grad
    params[1] -= lr * 2 * np.sum(linreg(batch_X, *params) - batch_y) / m

# 训练模型
lr = 0.001  # 学习率
num_epochs = 100  # 迭代次数
for epoch in range(num_epochs):
    for X, y in zip(X_train_scaled, y_train):  # 使用小批量或随机样本更新
        sgd([w, b], lr, [X], [y])

# 预测
y_train_pred = linreg(X_train_scaled, w, b)
y_test_pred = linreg(X_test_scaled, w, b)

# 输出训练误差和测试误差
print(f'Training MSE: {mean_squared_error(y_train, y_train_pred)}')
print(f'Testing MSE: {mean_squared_error(y_test, y_test_pred)}')
print(f'R^2 score on training set: {r2_score(y_train, y_train_pred)}')
print(f'R^2 score on testing set: {r2_score(y_test, y_test_pred)}')

请注意，上述代码中的`load_data`需要您根据自己实际情况来实现具体逻辑。此外，在实际应用时还需要考虑更多的细节如异常处理、更复杂的SGD实现（包括动态调整学习率等）。此代码仅供参考用途，请根据项目的具体需求调整和完善它。