基于大数据云平台的深度学习预测模型研究

Research on Deep Learning Prediction Model Based on Big Data Cloud Platform

近年来,随着云计算、大数据等技术的迅猛发展,如何快速、有效地从纷繁复杂的数据中获取有价值的信息成为当前大数据应用的关键问题。为此,对基于大数据云平台的深度学习预测模型进行研究,以对未来序列数据走势进行有效预测。首先对几种基于深度学习的长短序列预测模型进行对比分析,分析其与传统预测模型的区别及优势,提出一种加入dropout的轻量级GRU预测模型。采用代表性天气数据作为实验对象,实验结果表明,该方法的实验预测指标MAE(平均绝对误差)的平均值相比传统预测方法有所提高,从而有效验证了轻量级GRU预测方法的正确性与有效性。

In recent years,with the rapid development of cloud computing,big data and other technologies,how to quickly and effec⁃tively obtain valuable information from the complex data has become the key problem of the current big data development.Faced with the above problems,this paper uses deep learning model for prediction research,realizes in-depth analysis,integration and mining of potential information based on big data platform data base,and effectively forecasts the trend of future serial data.Specifically,this pa⁃per firstly makes a comparative analysis of several prediction models based on deep learning,analyzes the differences and advantages between them and traditional prediction models,and proposes a lightweight GRU prediction model with dropout.Based on the represen⁃tative weather data,the experimental prediction index are higher than the traditional prediction method.According to the experimental results,the correctness and effectiveness of the proposed lightweight GRU prediction method are verified.