移动大数据：模型到技术的路径指南

需积分: 10 175 浏览量更新于2024-07-18 收藏 8.17MB PDF 举报

身份认证购VIP最低享 7 折!

30元优惠券

资源详情

资源推荐

1.2 Big Data Analytics

The analysis of large data-sets in enterprises, the term of big data analytics is

associated with data science, business intelligence and business analytics. Data

science is deﬁned as a collection of fundamental principles that prom otes taking

information and knowledge from data [4]. Over the last years, data-driven

approaches like Business Intelligence (BI) and Business Analytics are characterized

indispensable to operating enterprises. BI is deﬁned as the methodologies , systems

and applications for collecting, preparing and analyzing data to provide information

helping decision makers. In other words, BI systems are data-driven decision

making systems [14], while Business Analytics are the techniques, technologies,

systems and applications that are used to analyze critical business data for sup-

porting them to understand their business environment and take business decisions

on time. The power of Business Analytics is to streamline vast amounts of data to

enhance its value, while BI mainly concentrates historical data in graphs and data

table reports as a way to provide answers to queries without streamlining data and

enhancing its value.

Business Analytics was commenced to outline the principal analytical element in

BI in the late 2000s. Afterwards, the terms of big data and big data analytics have

been utilised to describe analytical techniques for data- sets that are so large and

complex, needing advanced data storage, management, analysis and visualization

technologies. In that rapidly growing environment, the velocity of data makes the

conversion of data into valuable knowledge quickly a necessity. The differences

between conventional analytics and fast analytics with Big data are in analytics

characteristics (type, o bjective and method), data characteristics (type, age/ﬂow,

volume) and primary objective (Table 1)[15, 16].

The development of the Internet and later on the connectivity coming from the

web has contributed in the increase of the volume and speed of data. Since the early

2000s, Internet and Web technologies have been offering unique data collection and

Table 1 Conventional and big data analytics

Conventional analytics Big data analytics

Analytics

type

Descriptive, Predictive Predictive, Prescriptive

Analysis

methods

Hypothesis-based Machine learning

Primary

objective

Internal decision support and

performance management

Business processes driver and

data-driven Products

Data type Structured and deﬁned (formatted in

rows & columns)

Unstructured and undeﬁned

(unstructured formats)

Data

age/ﬂow

>24 h Static pool of data <Min Constant ﬂow of data

Data

volume

Tens of terabytes or less 100 terabytes to petabytes

8 K. Vassakis et al.

analysis for enterprises. Web 1.0 systems enable enterprises to establish a web

presence and offer their products/services online interacting with their customers.

Web 2.0 systems, including the introduction of social media networks like Face-

book, provide enterprises more data with information about enterprises, products

and customers. The ongoing increase of mobile devices against the number of

computers introduced a new era of business analytics, including the analysis of

user-generated content by social media channels. Mobile devices have the capa-

bility to promote e.g. highly mobile, location-aware and person-centered processes

and transactions. Therefore, Data-driven decision making is on data coming from

all the source s of enterprises, while predictions and machine learning are based on

traditional data and new innovative sources like IoT and AI.

Data analysis is the process of inspecting, cleaning, transforming and modeling

data gaining useful information for suggestions and support in decision-making. It

has multiple facets and approaches, encompassing diverse techniques under a

variety of names, in different business, science and social science schemes, while

“Big Data Analytics” refers to advanced analytic techniques, considering large and

various types of datasets to examine and extract knowledge from big data, con-

stituting a sub-process in gaining insights from big data process. Using advanced

technologies, Big Data Analytics (BDA) includes data management, open-source

programming like Hadoop, statistical analysis like sentiment and time-series anal-

ysis, visualization tools that help structure and connect data to uncover hidden

patterns, undiscovered correlations and other actionable insights.

The process of BDA is a resource for strategic decisions leading to signiﬁcant

improvements in operations performance, new revenue streams and competitive-

ness against rivals. In that context, the process of getting insights from big data can

be divided into two phases: data management and data analysis. Data management

is related with the processes and technologies for data generation, storage, mining

and preparation for analysis, while data analysis refers to the methods and tech-

niques for analysis and interpretation of the insights coming from big data [17]

(Fig. 3).

Analytics can be divided into four categories, ranging from descriptive and

diagnostic analytics to the more advanced predictive and prescr iptive analytics.

Fig. 3 Process of leveraging big data

Big Data Analytics: Applications, Prospects and Challenges 9

Descriptive analytics, based on historical and current data, is a signiﬁcant

source of insights about what happened in the past and the correlations between

various determinants identifying patterns using stat istical measures like mean, range

and standard deviation. Descriptive analytics using techniques like online analytical

processing (OLAP) exploits knowledge from the past experience to provide

answers in what’s happening in the organizations. Common examples of descrip-

tive analytics include data visualization, dashboards, reports, charts and graphs

presenting key metrics of enter prises including sales, orders, customers, ﬁnancial

performance etc.

Diagnostic analytics based also in historical data provide insights about the

root-cause of some outcomes of the past. Thus, organizations can take better

decisions avoiding errors and negative results of the past.

Predictive analytics is about forecasting and providing an estimation for the

probability of a future result, deﬁning opportunities or risks in the future. Using

various techniques including data mining, data modeling and machine learning, the

implementation of predictive analytics is signiﬁcant for any organization’s segment.

One of the most known applications of that type of analytics is the prediction of

customer behavior, determining operations, marketing and preventing risk. Using

historical and other available data, predictive analytics are able to uncover patterns

and identify relationships in data that can be used for forecasting [17]. Predictive

analytics in the digital era is a signiﬁcant weapon for organizations in the com-

petitive race. Therefore, organizations exploiting predictive analytics can identify

future trends and patterns, presenting innovative products/services and innovations

in their business models.

Prescriptive analytics provide a forecasting of the impact of future actions

before they are taken, answerin g “what might happen” as outcome of the organi-

zation’s actions. Therefore, the decision-making is improved taking under con-

sideration the prediction of future outcomes. Prescriptive analytics using high level

modeling tools is able to contribute remarkably to the performance and efﬁciency of

organizations, through smarter and faster decision with lower cost and risk and

identifying optimal solutions for resource allocation [18].

The advanced predictive and prescriptive analytics can play crucial role in

efﬁcient strategic decision making dealing with signiﬁcant problems of organiza-

tions like design and development of products/services, supply chain formation etc.

[19].

1.2.1 Big Data Analytics Applications

Nowadays, as the growing generation of available data is a recognized trend across

enterprises, countries and market segments, the majority of enterprises regardless

industry is collecting, storing and analyzing data in order to capture value. Digital

economy through the tremendous use of internet and digital services has trans-

formed almost all the industry sectors, including agriculture and manufacturing, to

more service-centered [20]. There are many and different sectors, like e-commerce,

10 K. Vassakis et al.

politics, science & technology, health, government services etc., where big data

analytics are applied. Data-driven companies from various industries clarify the

power of big data, making more accurate predictions leading on better decisions.

The large streams of data generated everyday need better infrastructures in order

to be captured, stored and analyzed. A market with a wide supply of new products

and tools designed to cover all the needs of big data has been created and it is

developing rapidly [ 21 ]. There is a wide variety of analytic tools that can be used to

perform BDA, among others on the basis of SQL queries, statistical analysis, data

mining, fast clustering, natural language processing, text analytics, data visualiza-

tion and artiﬁcial intelligence (AI). These techniques and tools provide easily and

rapidly exploitation of big data.

The knowledge derived from exploitation of big data provides enterprises added

value through new ways of productivity, growth, innovation and consumer surplus

[7], thus big data becomes a major determinant of competitiveness and enterprises

are in need of data analysis capacity to exploit the full potential of data.

Enterprises that learn to capitalize big data utilizing real-time information

coming from various sources like sensors, connected devices etc. can understand in

more detail their environment and deﬁne new trends, create new and innovative

products/services, respond quickly in changes and optimi ze their marketing actions.

The leverage of big data is able to contribute to the efﬁcient resources’ allocation

and supervision, waste reduction, facilitation of new insights and higher level of

transparency in different sections of enterprises from production to sales.

Therefore, BDA applications in almost every business sector exist. Applications

also in politics and e-government, science and technology, security and safety,

smart health and well-being exist [3]. In addition, there are plenty and v arious types

of big data applications among enterprises and industry sectors. BDA can be

employed in e-commerce and marketing applications like online advertisi ng and

cross-selling, while it helps enterprises to analyze customer behavior in shaping

360-degree customer proﬁle for implementation of targeted and optimized mar-

keting actions to impact customer acquisition and satisfaction. It offers better

understanding of customers’ behavior and preferences and thus improve customer

service.

Some examples of the ways BDA are exploited showing the signiﬁcance of

analytics in various themes [22]:

Marketing Market

basket

analysis

Recommendation

systems

Customer

Intelligence

Retention

modeling

Customer

churn

prediction

Processes Supply

chain

analytics

Demand and

supply forecasting

Business

Processes

analytics

Government Fraud

detection

Terrorism

Detection

Tax

avoidance

Cost

reduction

Social

security

(continued)

Big Data Analytics: Applications, Prospects and Challenges 11

(continued)

Risk

Management

Credit

risk

modeling

Market risk

modeling

Fraud

detection

Web and

Social media

Web

analytics

Social media

analytics

Multivariate

testing

Enterprises and organizations collect large amounts of security-relevant data

such as software application events, network events, people’s action events. The

generation of data coming from these actions are increasing rapidly per day as

organizations enable logging in more sources, running more software programs,

have more working employees and move to cloud solutions. Unfortunately, the

volume and variety of security data quickly become overwhelming and existing

analytical techniques cannot work efﬁciently and trustworthily. BDA applications

become part of security management and monitoring, since it contributes to

cleaning, preparation and analysis of various co mplex and heterogeneous datasets

efﬁciently [23]. One of the most common uses of BDA is fraud detection, thus

ﬁnancial institutions, governments and phone companies use big data technologies

to eliminate risk and enhance their efﬁcacy.

In addition, BDA is widely applied in supply chain and logistics operations

playing a signiﬁcant role in developing supply chain strategies and supply chain

operations management. BDA can support decision making through the under-

standing of changes of marketing conditions, identiﬁcation of supply chain risks

and exploiting supply chain capabilities to model innovative supply chain strate-

gies, thereby improving the ﬂexibility and proﬁtability of supply chain. BDA

contributes also in decision making at operational level, since it measures and

analyses supply chain performance taking into account demand planning, supplies,

production, inventory and logistics. It thus improves efﬁciency of operations,

measures supply chain performance, reduces process alterability and contributes to

the implementation of the best supply chain strategies at operational level [24].

Talking about digital and data-driven enterprises, the ﬁrsts coming in mind are

Google, Amazon, Apple and Facebook. Amazon that was born digital, exploited

big data achieving to disrupt traditional book market and became the leader in

digital shopping. Another example of a famous born-digital ﬁrm is Google that

harness data from engine search to digital marketing in order to provide and per-

sonalize search to its users, while Google and Facebook collect data providing

opportunities for personalized and customized marketing.

Nevertheless, traditional non-technological enterprises are also attempting to

gain data-driven beneﬁts. General Electric (GE) has developed a cloud-based

platform for Industrial Internet application named “Predix ” that provides real-time

insights for engineers to schedule maintenance checks, improves machine efﬁciency

and reduces downtime. GE this way provided new service value propositions in the

conservative market of the oil and gas industry, while it faces its most pressing

challenges: improving assets and operations productivity and eliminating the cost of

tacit knowledge from aging workforce [25].

12 K. Vassakis et al.

剩余345页未读，继续阅读

wang1062807258

粉丝: 13
资源: 272

移动大数据：模型到技术的路径指南

Requirements Engineering: A Roadmap

Software Architecture: a Roadmap文献阅读报告

Software Engineering for Real-Time: A Roadmap

SysSec红皮书：系统安全研究路线图The SysSec Red Book: A Roadmap for Systems Security Research

LTE Femtocell Roadmap from Concept to Reality

DataScientist-Roadmap:Análisisestadísticosexploratorios con python

4G Roadmap and Emerging Communication Technologies (Universal Personal Communications)

nlp-roadmap：ROADMAP（Mind Map）和KEYWORD面向有兴趣学习NLP的学生

data-science-roadmap::bullseye:个人路线图，以更深入地了解数据科学和相关领域

开发者成长之路：developer-roadmap

数据工程师学习路线图 / Roadmap to becoming a data engineer @ShowMeAI研究中心

data-science-roadmap:学习数据科学及相关领域的路线图

世界经济论坛：Enabling Measures Roadmap for Renuwable Hydrogen Europe

A roadmap for graphene

NLP：Roadmap for NLP涵盖NLP的理论知识、应用场景和工程实践等.zip

Agile.Adoption.Patterns.A.Roadmap.to.Organizational.Success.0321514521

Complete_data_scientist_roadmap:该存储库包含我遵循的成为数据科学家的完整路线图

OriseTech TFT Mobile Driver IC Roadmap: 2008 and Beyond

姚延栋：GreenplumDB 5.0 roadmap - 开源MPP数据库的稳步发展与敏捷开发策略

最新资源