Amazon EC2 Spot Price Prediction Using Regression Random Forests
Amazon EC2 Spot Price Prediction Using Regression Random Forests
复制标题
DOI：
10.1109/tcc.2017.2780159
复制
发表时间：
2020-01
影响因子：
6.5
通讯作者：
Veena Khandelwal;A. K. Chaturvedi;C. P. Gupta
中科院分区：
文献类型：
--
作者：
Veena Khandelwal;A. K. Chaturvedi;C. P. Gupta
Amazon EC2于2009年12月推出Spot实例，通过基于拍卖的市场机制出售其空闲容量。尽管价格极低，但云现货市场的利用率很低。现货定价是动态的，现货实例很容易出现出标失败。竞价复杂性是今天用户仍然害怕使用spot实例的另一个原因。本文提出了回归随机森林（RRFs）模型来预测未来一周和一天的现货价格。预测将帮助云用户提前计划何时获取现货实例，估计执行成本，并帮助他们进行投标决策，以最大限度地减少执行成本和投标失败概率。仿真与12个月的真实的亚马逊EC2现货历史跟踪预测未来现货价格显示所提出的技术的有效性。基于RRF的现货价格预测与现有的非参数机器学习模型的比较表明，基于RRF的预测精度优于其他模型。我们使用MAPE，MCPE，OOB错误和速度来衡量预测准确性。评估结果表明，对于66%至92%的提前一天预测，$MAPE MAPE=10%，而对于35%至81%的提前一天预测，预测时间小于1秒，$MCPE MCPE=15%。$MAPE MAPE=15%，用于71%到96%的一周前预测。
Spot instances were introduced by Amazon EC2 in December 2009 to sell its spare capacity through auction based market mechanism. Despite its extremely low prices, cloud spot market has low utilization. Spot pricing being dynamic, spot instances are prone to out-of bid failure. Bidding complexity is another reason why users today still fear using spot instances. This work aims to present Regression Random Forests (RRFs) model to predict one-week-ahead and one-day-ahead spot prices. The prediction would assist cloud users to plan in advance when to acquire spot instances, estimate execution costs, and also assist them in bid decision making to minimize execution costs and out-of-bid failure probability. Simulations with 12 months real Amazon EC2 spot history traces to forecast future spot prices show the effectiveness of the proposed technique. Comparison of RRFs based spot price forecasts with existing non-parametric machine learning models reveal that RRFs based forecast accuracy outperforms other models. We measure predictive accuracy using MAPE, MCPE, OOB Error and speed. Evaluation results show that $MAPE MAPE=10% for 66 to 92 percent and $MCPE MCPE=15% for 35 to 81 percent of one-day-ahead predictions with prediction time less than one second. $MAPE MAPE=15% for 71 to 96 percent of one-week-ahead predictions.