Chapter 1: Introduction
1.1 Research Background
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Chapter 1: Introduction
1.1 Research Background
1.1.1 Data-Driven Decision-Making
Data-driven decision-making is a process of transforming raw data into meaningful information and information into knowledge that can be applied to make decisions (Mandinach et al., 2006). The primary concept of data-driven decision-making involves making decisions based on data analysis rather than experience or intuition (Provost and Fawcett, 2013).
The world will undoubtedly become more data-driven, as data are very useful and valuable (Lohr, 2012). Rather than guessing, decision-making can become more data-driven in many areas, as data availability is greatly improved compared to the past (Jagadish et al., 2014). Organizations generate and gather large quantities of information in the course of everyday business that can be beneficial and valuable (Kabir and Carayannis, 2013). Due to the large scale of available data, companies in many industries are interested in using data to improve decisions and gain competitive advantages (Provost and Fawcett, 2013). From today’s perspective, decisions made without consulting data seem likely to fail (Shim et al., 2015), i.e., data-driven decisions are usually better decisions (McAfee et al., 2012).
1.1.2 Big Data
Due to technology advances, data is much more available than in the past (Jagadish et al., 2014). Data are becoming not only easier to access but also more amenable to computer processing as mathematical models and computer capabilities improve (Lohr, 2012). Advances in computer capabilities enable more sophisticated data analysis (Provost and Fawcett, 2013), resulting in a tendency towards data-driven discovery and Big Data-enabled decision-making (Lohr, 2012).
As many prior studies have reported, Big Data can be defined by the following properties in figure 1-1 (Ding et al., 2014; Gandomi and Haider, 2015; Kitchin and McArdle, 2016; Sivarajah et al., 2017). The volume relates to a large scale of datasets, a common challenge posed by Big Data (Sivarajah et al., 2017). Velocity relates to fast or real-time processing needed for quick decision-making, which is associated with the data collection speed, data transferring reliability, data storage efficiency, and excavation speed of discovering knowledge (Chen et al., 2015; Zhong et al., 2016).
Variety relates to the diversity of Big Data sources, which includes structured, semi-structured, and unstructured data (Gandomi and Haider, 2015; Kitchin and McArdle, 2016). Veracity relates to accuracy, truthfulness, and precision of data (Shim et al., 2015). Value relates to economic benefits of Big Data (Wamba et al., 2015). Viability
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
relates to making forecasts based on all relevant information (Fouad et al., 2015).
Visualization relates to reader-friendly presentation of data (Wamba et al., 2015).
Variability relates to inconsistencies in data flow (Katal et al., 2013).
Many studies defined that Big Data means having data on a scale that is too large and difficult to process efficiently by conventional analysis tools (Katal et al., 2013;
Madden, 2012; Provost and Fawcett, 2013). A survey made by IBM Institute for Business Value and Saïd Business School at the University of Oxford found out that more than half of respondents considered that Big Data is datasets that bigger than one terabyte (Schroeck et al., 2012). Another survey made by Intel IT center defined that storing more than ten terabyte of dataset as Big Data (Intel IT center, 2012).
Besides, the volume of Big Data can also be measured by the number of records, transactions, tables, or files; an interviewee of TDWI survey indicated that their company never counted how many terabytes their data is, but they processed billions of data record (Russom, 2014).
In addition to volume, the variety of data makes Big Data really big (Russom, 2014). The Method for an Integrated Knowledge Environment (MIKE2.0) project introduced a definition that Big Data relates to big complexity, which means that even though the volume is not very large, data can be considered as Big Data as long as the data is complex and unstructured, such as videos (Rindler et al., 2013).
The recent interest in data-driven insights has contributed to the popularity of Big Data (Madden, 2012). While it might be a marketing term, it also represents a new approach to exploring the world and making decisions (Lohr, 2012). The present is a new age of data-driven decisions and Big Data, with increasingly more decisions of many businesses relying on data analysis to improve efficiency and performance (Jagadish et al., 2014). Big Data opens great opportunities for real-time decision-making, operational and customer intelligence, business innovation, and gaining competitive advantages (Chen et al., 2015). However, it is essential to ensure that humans can fully comprehend the analysis results and avoid getting lost in the data sea (Jagadish et al., 2014).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Figure 1-1 Properties of Big Data
1.1.3 Big Data Application in Industries
Financial Industry
In financial industry, by not only analyzing structured data from exchanges, banks, and data vendors, but also analyzing unstructured data from news, twitters, and social media, Big Data enables financial firms to discover innovative and strategic ways to manage portfolio, analyze risk exposure, perform enterprise-level analytics, and comply with regulations (Fang and Zhang, 2016). For example, Big Data technology enabled Bank of America to analyze all transactional data, customer data across multiple channels and relationships, and unstructured data at once, which successfully improve the understanding about customers (Davenport and Dyché, 2013). Give another example of a European company that provides credit-scoring solution for banks and consumer lending companies, the company used Big Data to analyze loan payment data and Facebook user data, and improved their scoring models, saved money on credit losses, expanded potential client base, and even increased revenue (Fang and Zhang, 2016). According to the study of Davenport and Dyché (2013), banks such as Wells Fargo, Bank of America, and Discover used Big Data to analyze billions records of unstructured and semi-structured data including data from financial industry and other industries, such as website clicks, transaction records, bankers’ notes, and voice recordings from call centers. Big Data analytics help them improve their customer relationship, enhance the quality of customer interaction, and segment their customer better (Davenport and Dyché, 2013).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Healthcare Industry
In Healthcare industry, the amount of available health data is increasing because of medical applications, remote monitoring technologies, and biological sensors; and these data usually need to be processed in real-time to not delay treating (Ong et al., 2016). By analyzing various data such as health record data, sensors' data, patient arrangement data, visualizing record data, Big Data analytics is helpful for disease interpreting, diseases avoiding, and diseases curing in healthcare industry (Das et al., 2018). For example, Big Data analytics enables United Healthcare to analyze unstructured data such as the recorded voice files from customer calls to call center, and understanding its customer satisfaction better (Davenport and Dyché, 2013). In another example, the study of Chen et al. (2017) analyzed more than 20 million records of inpatient department data. The inpatient department data includes structured data such as laboratory data and patient’s basic information, and it includes unstructured data such as patient’s narration of illness, doctor’s interrogation record and diagnosis. For complex diseases, analyzing both structured data and unstructured data improved the feature descriptions of diseases and the accuracy rate of disease risk prediction, which is helpful for disease preventing and early treatment (Chen et al., 2017).
Retail Industry
In retail industry, Big Data may help companies to evaluate customers’ behavior, segment and target customers more accurately, and forecast demand trends of products (Srivastava, 2018). For example, Walmart and Kohl’s improved product selection and price markdown times by analyzing sales, pricing, economic, demographic, and weather data (Lohr, 2012). Another example is that Big Data enabled Sears to process petabytes of customer, product, sales, and campaign data in real-time; and resulted in understanding increase marketing return, increasing customer return rate, and decreasing the release time of a set of complex marketing campaigns (Davenport and Dyché, 2013). Moreover, in the case study of Evans and Kitchin (2018), a large retail store operating in Ireland analyzed large amount of data in its ERP system, SCM system, CRM system, stock control system, online order fulfilment system, and so on. Big Data is found out to be helpful for decision- making, increasing efficiencies, improving customer service, reducing risk and costs, and managing the work of staff (Evans and Kitchin, 2018).
Manufacturing Industry
In manufacturing industry, innovative applications, such as active preventive maintenance, production line optimization, and energy consumption optimization, can be motivated by analyzing device data, product data, and command data (Wan et al.,
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2017). For example, a sensor of GE can generate 500 gigabytes of data per day, and analyzing the data from more than ten thousands of sensors helped GE optimize inspection, maintenance and repair processes of machines, which improved the efficiency and increased the revenue (Davenport and Dyché, 2013). Another example is that Intel used Big Data to decrease its quality checking time of chips by analyzing their historical data from pre-release chips, and successfully launching its chips in the market faster and decreasing its quality ensuring cost (“How big data analytics yields big gains,” 2017). As a final illustration, according to the study of Zhang et al. (2017), by analyzing real-time data such as flow rate, pressure, temperature, vibration signal, as well as non-real-time data such as maintenance history and failure record, Big Data enabled a cleaner production manufacturing company to provide better products, customized products, better service, and more innovative service. With these improvements, the company successfully decreased cost, improved manufacturing efficiency, increased revenue, increased customer satisfaction, increased customer loyalty, identified more potential customers, innovated its business model, and improved strategic cooperation with other companies (Zhang et al., 2017).
Service Industry
In service industry, Caesars Entertainment implemented Big Data tools to improve marketing and service in real time by analyzing data about its customers from its Total Rewards loyalty program, web clickstreams, and from real-time play in slot machines (Davenport and Dyché, 2013). Furthermore, Starwood Hotels and Resorts used Big Data to improve its pricing strategy and demand forecasting and it successfully increased their revenue by analyzing their structured data (such as bookings data and cancellations data) and external semi-structured data (such as weather reports) (“How big data analytics yields big gains,” 2017).