波士顿房价预测：数据预处理及线性回归模型构建

{ "title": "导入所需软件包", "description": "本示例展示了如何使用 Python 进行波士顿房价预测，包括数据预处理、线性回归模型训练、模型保存及预测等步骤。", "keywords": "波士顿房价预测, 线性回归, 数据预处理, 异常值处理, 模型训练, 模型保存, Python", "content": "import os\nimport numpy as np\nimport pandas as pd\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import MinMaxScaler\nfrom sklearn.linear_model import LinearRegression\nfrom sklearn.metrics import mean_squared_error\n\ndata = pd.read_csv("./boston_house_prices.csv")\n\n预览前5行数据\ndata.head()\n\n箱线图查看异常值分布\nimport matplotlib.pyplot as plt\n\ndef draw_boxplot(data):\n# 绘制除了 CHAS 属性之外的每个属性的箱线图\ndata_col = list(data.columns)\ndata_col.remove("CHAS")\n\n复制代码\n# plt.figure(figsize=(2, 9), dpi=400)\nplt.figure(figsize=(5, 5), dpi=300)\nplt.subplots_adjust(wspace=0.6)\nfor i, col_name in enumerate(data_col):\n plt.subplot(3, 5, i + 1)\n plt.boxplot(data[col_name],\n showmeans=True,\n whiskerprops={"color": "g", "linewidth": 0.4, 'linestyle': "--"},\n flierprops={"markersize": 0.4},\n meanprops={"markersize": 1})\n plt.title(col_name, fontdict={"size": 5}, pad=2)\n plt.yticks(fontsize=4, rotation=90)\n plt.tick_params(pad=0.5)\n plt.xticks([])\n\nplt.show()\ndraw_boxplot(data)\n\n四分位处理异常值\nnum_features = data.select_dtypes(exclude=['object', 'bool']).columns.tolist()\nfor feature in num_features:\nQ1 = data[feature].quantile(q=0.25)\nQ3 = data[feature].quantile(q=0.75)\nIQR = Q3 - Q1\ntop = Q3 + 1.5 * IQR\nbot = Q1 - 1.5 * IQR\nvalues = data[feature].values\nvalues[values > top] = top\nvalues[values < bot] = bot\ndata[feature] = values.astype(data[feature].dtypes)\n\n再次查看箱线图，异常值已被临界值替换\ndraw_boxplot(data)\n\nX = data.drop(['MEDV'], axis=1)\ny = data['MEDV']\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)\n\nscaler = MinMaxScaler()\nx_train = scaler.fit_transform(X_train)\nx_test = scaler.transform(X_test)\n\nlr_model = LinearRegression()\nlr_model.fit(X_train, y_train)\n\ny_pred = lr_model.predict(X_test)\nmse = mean_squared_error(y_test, y_pred)\nprint('MSE:', np.sqrt(mse))\n\nfrom joblib import dump\nimport shutil\n\nmodel_path = "./linear_model"\n\nif os.path.exists(model_path):\nshutil.rmtree(model_path)\ndump(lr_model, model_path)\n\nfrom joblib import load\n\nlr_model = load('linear_model')\nlr_model.predict(X_test[:1])内容：array([28.9563193])