Organisations are increasingly gene

Organisations are increasingly generating large volumes of data
as result of instrumented business processes, monitoring of user
activity [14,127], web site tracking, sensors, finance, accounting,
among other reasons. With the advent of social network Web sites,
users create records of their lives by daily posting details of activities they perform, events they attend, places they visit, pictures they take, and things they enjoy and want. This data deluge
is often referred to as Big Data [99,55,17]; a term that conveys the
challenges it poses on existing infrastructure with respect to storage, management, interoperability, governance, and analysis of the
data.
In today’s competitive market, being able to explore data to understand customer behaviour, segment customer base, offer customised services, and gain insights from data provided by multiple
sources is key to competitive advantage. Although decision makers
would like to base their decisions and actions on insights gained
from this data [43], making sense of data, extracting non obvious
patterns, and using these patterns to predict future behaviour are
not new topics. Knowledge Discovery in Data (KDD) [50] aims to
extract non obvious information using careful and detailed analysis and interpretation. Data mining [133,84], more specifically,
aims to discover previously unknown interrelations among apparently unrelated attributes of data sets by applying methods from
several areas including machine learning, database systems, and
statistics. Analytics comprises techniques of KDD, data mining, text
mining, statistical and quantitative analysis, explanatory and predictive models, and advanced and interactive visualisation to drive
decisions and actions [43,42,63].
Fig. 1 depicts the common phases of a traditional analytics workflow for Big Data. Data from various sources, including
databases, streams, marts, and data warehouses, are used to build
models. The large volume and different types of the data can demand pre-processing tasks for integrating the data, cleaning it, and
filtering it. The prepared data is used to train a model and to estimate its parameters. Once the model is estimated, it should be validated before its consumption. Normally this phase requires the use
of the original input data and specific methods to validate the created model. Finally, the model is consumed and applied to data as
it arrives. This phase, called model scoring, is used to generate predictions, prescriptions, and recommendations. The results are interpreted and evaluated, used to generate new models or calibrate
existing ones, or are integrated to pre-processed data.
Analytics solutions can be classified as descriptive, predictive,
or prescriptive as illustrated in Fig. 2. Descriptive analytics
uses historical data to identify patterns and create management
reports; it is concerned with modelling past behaviour. Predictive
analytics attempts to predict the future by analysing current and
historical data. Prescriptive solutions assist analysts in decisions by
determining actions and assessing their impact regarding business
objectives, requirements, and constraints.
Despite the hype about it, using analytics is still a labour intensive endeavour. This is because current solutions for analytics are
often based on proprietary appliances or software systems built for
general purposes. Thus, significant effort is needed to tailor such
solutions to the specific needs of the organisation, which includes
integrating different data sources and deploying the software on
the company’s hardware (or, in the case of appliances, integrating the appliance hardware with the rest of the company’s systems) [120]. Such solutions are usually developed and hosted on
the customer’s premises, are generally complex, and their operations can take hours to execute. Cloud computing provides an interesting model for analytics, where solutions can be hosted on
the Cloud and consumed by customers in a pay-as-you-go fashion.
For this delivery model to become reality, however, several technical issues must be addressed, such as data management, tuning
of models, privacy, data quality, and data currency.
This work highlights technical issues and surveys existing work
on solutions to provide analytics capabilities for Big Data on the
Cloud. Considering the traditional analytics workflow presented in
Fig. 1, we focus on key issues in the phases of an analytics solution.
With Big Data it is evident that many of the challenges of Cloud
analytics concern data management, integration, and processing.
Previous work has focused on issues such as data formats, data
representation, storage, access, privacy, and data quality. Section 3
presents existing work addressing these challenges on Cloud environments. In Section 4, we elaborate on existing models to provide
and evaluate data models on the Cloud. Section 5 describes solutions for data visualisation and customer interaction with analytics solutions provided by a Cloud. We also highlight some of the
business challenges posed by this delivery model when we discuss
service structures, service level agreements, and business models.
Security is certainly a key challenge for hosting analytics solutions
on public Clouds. We consider, however, that security is an extensive topic and would hence deserve a study of its own. Therefore,
security and evaluation of data correctness [130] are out of scope
of this survey.