编号：C204
大小：17M
环境：Jupyter Notebook
简介：基于LSTM单变量预测_电力消耗预测
案例实现思路：
1.加载数据集、预处理
2.特征工程
3.构建模型
4.模型编译、训练、验证
5.模型.测试
6.结果可视化

运行展示

可视化显示DOM_MW的数据分布情况

可视化显示DOM_MW的数据分布情况

显示模型结构

模型训练
Epoch 1/10
409/409 [==============================] – 3s 3ms/step – loss: 0.1614 – val_loss: 0.0840
Epoch 2/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0721 – val_loss: 0.0686
Epoch 3/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0604 – val_loss: 0.0564
Epoch 4/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0486 – val_loss: 0.0423
Epoch 5/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0336 – val_loss: 0.0307
Epoch 6/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0255 – val_loss: 0.0249
Epoch 7/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0219 – val_loss: 0.0224
Epoch 8/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0197 – val_loss: 0.0216
Epoch 9/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0181 – val_loss: 0.0187
Epoch 10/10
409/409 [==============================] – 1s 2ms/step – loss: 0.0168 – val_loss: 0.0165
显示 train loss 和 val loss

绘制模型验证结果

绘制test中前100个点的真值与预测值

配套文件

我们提供完整项目文件清单如下：
文件目录
├ 1.项目源码
├ 2.运行截图
└ 3.演示视频