Estimating Cloud Spot Instances for Elastic Cloud Environments

Cloud services are used globally in different kinds of applications. In order to understand the price variation for spot instance prices the prediction algorithms are required. In this context the proposed work introduces based on data driven technique price prediction model. The data mining technique provides us the ability to learn with the historical data trends and identify or recognize the similar trends of the data. This ability makes it essential for various new generation applications. The proposed method is a supervised learning model and implemented using the SVR (support vector regression) model. The proposed method works in three phases. In first phase, the data collection and preprocessing is adopted. In next phase, the sliding window protocol is executed for grouping of similar data patterns. Finally, the training and testing of SVR model is performed for predicting the price of spot instances. The performance of the proposed system is evaluated in terms of the prediction accuracy.