
QRLSTM-based recurrent neural networks for quantile regression interval prediction with complete Matlab source codes
5星
- 浏览量: 0
- 大小:None
- 文件类型:ZIP
简介:
QRLSTM LSTM recurrent neural networks accompanied by quantile regression interval prediction, featuring a complete MATLAB source code package. Recommended running environment: MATLAB 2023 or higher versions.
The models performance metrics are evaluated based on the provided dataset:
1. Mean Squared Error (MSE): 82.7315
Root Mean Squared Error (RMSE): 9.0957
Average Absolute Error (MAE): 6.1943
Average Relative Error (ARE): 76.8966%
R-squared coefficient: 97.5857%
QRLSTM模型在训练集上的误差指标分析如下:
均值平方差(MSE)数值为65.7956;
平方根平均值(RMSE)数值为8.1115;
平均绝对偏差(MAE)数值为4.7284;
平均相对百分比误差(MAPE)数值为107.0846%;
决定系数值达到95.8516%。该训练集的空间覆盖率值为:0.88413,其平均宽度占比率是:0.20854。该测试集的空间覆盖率值为:0.90452,其平均宽度占比率是:0.19676。
全部评论 (0)
还没有任何评论哟~


