update some readme

Author: pinard.liu

classic-machine-learning/regression_production_example.ipynb

@@ -5,27 +5,49 @@
    "metadata": {},
    "source": [
     "<h3>可用于产品的线性回归算法建流程</h3>\n",
-    "1. 载入数据集\n",
-    "2. 数据预览\n",
-    "3. 获取特征数据和输出数据\n",
-    "4. 缺失值处理，异常值处理（若有）\n",
-    "5. 划分数据集为训练集和验证集\n",
-    "6. 训练集数据的标准化\n",
-    "7. 使用不同的参数在训练集上训练出模型(使用训练集交叉验证是可选的)\n",
-    "8. 观察训练集的MSE和RMSE\n",
-    "9. 验证集数据的标准化\n",
-    "10. 使用训练出的模型在验证集上做预测，并观察验证集的MSE和RMSE\n",
-    "11. 画图观察(可选)\n",
-    "12. 重复步骤7-11，选择验证集上MSE或RMSE最小的模型作为模型输出\n",
-    "13. 保存模型，加载模型做预测,共3种方法\n",
+    "<h4>数据准备阶段</h4>\n",
+    "\n",
+    "1.载入数据集\n",
+    "\n",
+    "2.数据预览\n",
+    "\n",
+    "3.获取特征数据和输出数据\n",
+    "\n",
+    "<h4>特征工程阶段</h4>\n",
+    "\n",
+    "4.缺失值处理，异常值处理，高级特征生成（若有）\n",
+    "\n",
+    "5.划分数据集为训练集和验证集\n",
+    "\n",
+    "6.训练集数据的标准化\n",
+    "\n",
+    "<h4>模型训练阶段</h4>\n",
+    "\n",
+    "7.使用不同的参数在训练集上训练出模型(使用训练集交叉验证是可选的)\n",
+    "\n",
+    "8.观察训练集的MSE和RMSE\n",
+    "\n",
+    "<h4>模型验证阶段</h4>\n",
+    "\n",
+    "9.验证集数据的标准化\n",
+    "\n",
+    "10.使用训练出的模型在验证集上做预测，并观察验证集的MSE和RMSE\n",
+    "\n",
+    "11.画图观察(可选)\n",
+    "\n",
+    "12.重复步骤7-11，选择验证集上MSE或RMSE最小的模型作为模型输出\n",
+    "\n",
+    "<h4>算法上线阶段</h4>\n",
+    "\n",
+    "13.保存模型，加载模型做预测,共3种方法\n",
     "    * 使用python pickle API 保存\n",
     "    * 使用python 加载pickle文件预测\n",
     "    * 使用sklearn joblib API 保存\n",
     "    * 使用sklearn joblib API 加载sklearn模型预测\n",
     "    * 保存为PMML文件\n",
     "    * 使用PMML Java API加载模型预测\n",
     "\n",
-    "<h5>by Pinard Liu</h5>"
+    "<h5>by Pinard Liu @ 20190131</h5>"
    ]
   },
   {