Article Preview
TopIntroduction
The volume and variety of relevant business data sources have been proliferating wildly worldwide in combination with most organizations having relatively limited ability to integrate cross-organizations systems operations into a cohesive company-wide information management resource. This massive increases in data volumes and variety has been named Big Data (BD). It has created opportunities and problems for company efforts to enhance IT decision support and its value to business managers in many organizations. For example, in the healthcare sector, interest in Big Data Analytics (BDA) has become widespread in academic and practitioner circles because the potential benefits are great and the consequences of managing BD poorly have led to unnecessary increases in medical costs, including wasted time for both patients and service providers (Bodenheimer, 2005; Grantmakers In Health, 2012; Herrick et al., 2010; The Kaiser Family Foundation, 2012).
The large data volumes being generated daily at unprecedented speeds from heterogeneous sources (e.g., health, government, social networks, marketing, financial) arise from the many technological trends such as the Internet and Cloud Computing (Botta et al., 2016), as well the widespread use of smart devices. Unlike traditional data, BD large growing data sets include heterogeneous formats: structured, unstructured and semi-structured data which provides the raw material essential for Big Data Analytics (BDA). Adding to the burden of effectively managing BDA, the increasingly complex nature of BD requires relatively more powerful technologies (equipment, tools, techniques, skill sets, and relatively more sophisticated algorithms).
What Is BDA and Why?
Potentially BDA promises to address such problems by enabling organizations to analyze large volumes and wide variety of data collected across a wide range of networks to effectively support evidence-based actionable decision making (Raghupathi & Raghupathi, 2014, Watson, 2014). In their efforts to better understand the more complex nature of BD, many data scientists (Emani et al. 2015; Furht & Villanustre, 2016; Gandomi & Haider 2015) have identified several of its basic characteristics: Volume accumulated from millions of devices and applications McAfee et al. (2012), which International Data Corporation estimated at 4.4 Zettabytes (ZB) in 2013, doubling every 2 years to 8 ZB in 2015 (Rajaraman, 2016), and estimated 40 ZB by 2020 (Kune et al., 2016). Velocity with which the data is generated and processed for producing timely actionable conclusions in various demanding applications. Variety of data from distributed sources in multiple formats including videos, documents, commentary, blue prints, and logs, with data sets consisting of structured and unstructured data, public or private, local or distant, shared or confidential, with various degrees of completeness, precision, etc. Verification where processed data has to conform to prior specifications. Validation where the purpose of the data meets prior criteria and is accepted as fulfilled. Complexity of the data due to the number of relationships which may change over time.
Besides BD, BDA encompasses an increasing arsenal of analytical techniques such as descriptive and prescriptive tools designed for analyzing a large proportion of text-based documents and other unstructured data. For example, in the health care area we might have physician's written notes, prescriptions, and medical imaging (Groves et al., 2013). New database management systems such as MongoDB, MarkLogic and Apache Cassandra can be used for data integration, retrieval, transferring data between different operating systems. To manage massive data volumes in various formats, various systems like Apache HBase and NoSQL can be used. These BDA tools have sophisticated functionalities which facilitate data integration potentially able to provide useful business insights (Jiang et al., 2014; Murdoch and Detsky, 2013; Wang et al., 2015). With BDA powerful systems and distributed applications supporting an increasing number of such systems in a wide variety of business sectors such as smart grid systems (Chen et al., 2014a), healthcare systems (Kankanhalli et al., 2016), retailing systems (Schmarzo, 2013), government systems (Stoianov et al., 2015), and many others.