摘要: |
水稻是我国主要的粮食作物,水稻产量对我国粮食安全有着重要的影响,水稻产量的预测对我国农业生产具有重要意义。传统方法采用时间序列及横截面数据基于统计和机器学习的方法对水稻产量进行预测,但缺乏对多种方法的对比分析。从时间序列预测和横截面数据预测两种角度,利用ARIMA、LSTM、SVR、MLP这4种模型,通过吉林省水稻产量、病虫害及其他特征历史数据对吉林省水稻产量进行预测,并对不同模型的预测结果进行了对比分析。 |
关键词: 水稻产量 ARIMA LSTM SVR MLP |
DOI: |
分类号: |
基金项目:农业部2017年创新人才项目;中国农业科学院科技创新工程项目(CAAS-ASTTP-2017-AII-02) |
|
Comparative Analysis of Rice Yield Forecasting Based on Time Series Analysis and Cross-sectional Prediction in Jilin Province of China |
chenwei1, qiweiyan1, yuanfuxiang2, lizhemin1
|
1.Agricultural Information Institute of CAAS;2.Jilin Institute of Meteorological Sciences
|
Abstract: |
Rice is the main food crop in China. Rice yield has an important impact on China’s food security. Rice yield prediction is of great significance to China’s agricultural production. Traditional methods use time series based on statistical methods and cross-sectional data based on machine learning methods to predict rice yield. However, there is a lack of comparative analysis of multiple methods. In this paper, we propose to use historical rice yield, disease, pest and other characteristics data to predict rice yield in Jilin province, China. Four models are used to predict rice yield. ARIMA and LSTM are utilized for time series prediction. Cross-sectional prediction is also conducted using SVR and MLP. The prediction results of different models are analyzed and compared. |
Key words: rice yield, ARIMA, LSTM, SVR, MLP |