基于注意力机制的区域小学入学规模预测

学龄人口是区域教育资源配置的重要依据，对区域内下一年小学入学规模进行准确预测，可以为区域内教育管理部门对教育资源进行调配提供辅助决策支持.该文针对区域内小学入学规模预测问题，考虑区域经济、人口等相关因素和小学入学规模的关联关系，提出了基于注意力机制的循环网络预测模型.该模型以长短时记忆网络模型为基础，引入注意力机制，自动提取小学入学规模与经济、人口等特征之间的关联关系以及进一步增强历史关键时间点的信息表达，提升预测准确率.在采用真实数据集进行试验的结果说明，该模型对比其它模型在多个评价指标上均有提升，具有更准确和更稳定的预测效果.

Prediction of regional primary school enrollment scale based on attention mechanism

The school-age population is an important basis for the allocation of educational resources in a region. An accurate prediction of the enrollment scale of the primary schools for the next year in the region can provide auxiliary decision support for the allocation of educational resources by the education and management departments in the region. In order to predict the enrollment scale of primary schools in a region， a circular network prediction model is proposed based on the attention mechanism which considered the correlation between regional economy， population and the enrollment scale of primary schools. This model is based on the Long Short-time Memory network model， and introduces the attention mechanism to automatically extract the correlation the primary school enrollment scale and the characteristics like economy and population， and further enhance the information expression of critical historical moments to improve the prediction accuracy. Using real data sets，the test results show thatthis model has improved in many evaluation indexes， and has a more accurate and stable prediction effect compared with other models.