1: Data Mining in Cloud Scheduling Cloud computing hosts application in public or private cloud. Cloud computing promises to deliver two main objectives to end users: cost efficiency and flexibility. Cloud schedulers allocates VM’s on the same machine to increase the system utilization.
When VM’s is scheduled, they do share hardware resources like last level cache and network switch. This compromises security and privacy drawbacks because resource isolation isn’t highly imposed. Like, when memory capacity is partitioned conflict in last level cache will still leak the information about the co-scheduled programs.
There is related work regarding this topic like denial of service, side-cannel attacks, denial of service and data leakage in cloud services. Data mining can determined the co scheduled victim application against big dataset of past seen workloads.2: Electricity Budget Classification and Demand Side Management Using Data Mining Predicting electricity prices has important role in making decisions in electricity market. Especially in Pakistan due to the shortage of energy, there is need to consider the Demand Side Management to fulfill people demand regarding their use of electricity. Demand Side Management, is the alteration of customer demand for energy by different methods such as economic motivations and behavioral revolution through education. In some application just like Demand Side Management, decisions can be made on various price thresholds.
It can be done to achieve classes of future prices, which can be used to cast electricity prices classification problem. The effectiveness of different data mining methods for electricity price classification and thresholding can be seen.3: Educational Data Mining (EDM) Educational institutions collect and process the large amount of data, such as student attendance record, enrolment, and their examination results. Mining this data helps to find information. Fast growth in educational data tells that extracting these big amounts of data needs a more advanced set of algorithms.
Old data mining algorithms cannot be implemented to educational problems because they have a special purpose. This means a preprocessing algorithm will be modified first and then specified data mining algorithm can be applied then. Clustering is the one data-mining algorithm for EDM. The International Educational Data Mining Society elaborates the EDM as an emerging discipline, concerned with improving methods for examining the unusual kinds of data that arise from educational backgrounds, and applying those methods to better know students, and the environments which they learn in.