MSE 值比应有的要大得多

Question

知道我在这里做错了什么吗： 我的数据集约为 20k 行，mse 约为 11298955095.811989，我不太确定我做错了什么？ 我正在尝试查找哪个数据集以及哪个 k 给出最小值，但没有一个值有任何意义：

import random
def split_df(dataframe):
    data_rows = dataframe.values.tolist()

    random.shuffle(data_rows)

    train_val = 0.7
    split_index = int(len(data_rows) * train_val)

    train_data = data_rows[:split_index]
    test_data = data_rows[split_index:]

    train_df = pd.DataFrame(train_data, columns=dataframe.columns)
    test_df = pd.DataFrame(test_data, columns=dataframe.columns)

    return train_df, test_df

from sklearn.neighbors import KNeighborsRegressor
from sklearn.metrics import mean_squared_error

dataframes = [no_nulls, outliers_removed, mean_imputed, median_imputed]
target = 'median_house_value'
k_vals = range(1, 30)

for df in dataframes:
    mse_dictionary = {}
    X = df.drop(columns=[target])  # Features
    y = df[target]  # Target variable

    # Split data into train and test
    train_df, test_df = split_df(df)

    # Separate features and target variable for train and test
    X_train = train_df.drop(columns=[target])
    y_train = train_df[target]
    X_test = test_df.drop(columns=[target])
    y_test = test_df[target]
    
    for k in k_vals:
        knn_regressor = KNeighborsRegressor(n_neighbors=k) 
        knn_regressor.fit(X_train, y_train)
        predictions = knn_regressor.predict(X_test)
        squared_errors = (predictions - y_test) ** 2  # Calculate squared errors
        mse = squared_errors.mean()  # Compute mean of squared errors to get MSE
        mse_dictionary[k] = mse

    print(f"MSE Dictionary for dataframe {df}: {mse_dictionary}")
    # Additional information for debugging
    print(f"Max MSE: {max(mse_dictionary.values())}")
    print(f"Min MSE: {min(mse_dictionary.values())}")
    print(f"Mean MSE: {sum(mse_dictionary.values()) / len(mse_dictionary)}")

Answer 1

事实证明这些值是正确的，回归模型对于小数据帧来说不太准确。