| 赵成兵,刘丹秀,谢新平,刘静.基于时间序列的季节性气温变化预测研究[J].安徽建筑大学学报,2022,30(): |
| 基于时间序列的季节性气温变化预测研究 |
| Research on Temperature Time Series Based on Seasonality |
| 投稿时间：2021-04-07 修订日期：2021-05-24 |
| DOI： |
| 中文关键词: 随机森林 自回归移动平均模型 平均温度 季节性 Python |
| 英文关键词: random forest autoregressive moving average model average temperature seasonality Python |
| 基金项目:安徽省高等学校自然科学基金重点项目,KJ2020A0479,KJ2018A0517 |
|
| 摘要点击次数: 1757 |
| 全文下载次数: 512 |
| 中文摘要: |
| 气温的动态结构受降水量等因素的影响，可以通过这些因素对气温时间序列进行分析。在Python语言环境下，提取数据中的季节性信息进行One-Hot编码，并作为随机森林模型的输入特征，对月平均气温进行拟合与预测。最后利用随机搜索和网格搜索对拟合模型中的超参数进行优化。结果表明：考虑季节性的随机森林模型拟合效果优于简单随机森林模型，预测数据变化趋势与实际观测基本一致，拟合精度可以达到96.14%；在两种方法对超参数优化之后，模型拟合精度可以达到96.45%。 |
| 英文摘要: |
| The dynamic structure of temperature is affected by precipitation and other factors, which can be used to analyze temperature time series. The seasonal information in the data is extracted with one-hot encoding by python language, and used as the input characteristic of the random forest model to fit and forecast the monthly average temperature. Finally, the super parameters in the fitting model are optimized by using random search and grid search. The results show that: the effect of the seasonal random forest model is better than that the simple random forest model, the trend of the predicted data is basically consistent with the actual trend, and the fitting accuracy can reach 96.14 percent; after the optimization is finished, the accuracy of the model can be up to 96.45 percent. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
