No abstract

Clustering is used as an important procedure in the process of data mining, where information of large datasets are transformed into meaningful and concise data. It performs activities like pattern representation, using of clustering algorithms and their validation, data abstraction and finally result generated. Clustering has many categories of algorithms such as partition-based, hierarchical-based, density-based, grid-based etc. Partition-based is the centroid-based clustering. Hierarchical-based clustering is link-based. Density-based is clustering is focused on area of higher density in the dataset. Grid-based clustering relies on size of the grid. In this paper, we discussed different clustering techniques as well as, a detailed review on the partition-based and hierarchical-based algorithms. Finally we compare clustering algorithms on the basis of attributes like time complexity, capacity of handling large datasets, scalability, sensitivity to outliers and noise, and also discussed result after solving a particular dataset implemented in cloud computing environment.

No abstract

PurposeIn modern cloud services, resource provisioning and allocation are significant for assigning the available resources in efficient way. Resource management in cloud becomes challenging due to high energy consumption at data center (DC), virtual machine (VM) migration, high operational cost and overhead on DC.Design/methodology/approachIn this paper, the authors proposed software-defined networking (SDN)-enabled cloud for resource management to reduce energy consumption in DC. SDN-cloud comprises four phases: (1) user authentication, (2) service-level agreement (SLA) constraints, (3) cloud interceder and (4) SDN-controller.FindingsResource management is significant for reducing power consumption in CDs that is based on scheduling, VM placement, with Quality of Service (QoS) requirements.Research limitations/implicationsThe main goal is to utilize the resources energy effectively for reducing power consumption in cloud environment. This method effectively increases the user service rate and reduces the unnecessary migration process.Originality/valueAs a result, the authors show a significant reduction in energy consumption by 20 KWh as well as over 60% power consumption in the presence of 500 VMs. In future, the authors have planned to concentrate the issues on resource failure and also SLA violation rate with respect to number of resources will be decreased.

Forecasting is a technique commonly used in the study of time series to forecast a variable response for a specified period of time, such as monthly earnings, stock performance, or unemployment figures. Forecasting is historical data behavior to determine the direction of future trends. Therefore, many machine learning algorithms are used in the past few years. In this study, a summary of an extreme learning machine with MapReduce technique (ELM_MapReduce) is presented. This technique is based on the concept of processing large amount of historical data and application of extreme learning machine to achieve fast learning speed. As stock market data is large set of historical data that need time to process, MapReduce method is used to handle such limitations. The technique shows the advantages and disadvantages of using MapReduce method in ELM and can be used in different areas of research.