COIT20253 Business Intelligence using Big Data : Assessment 3

USE OF BIG DATA TECHNOLOGIES

1 Introduction

The contemporary era, often known as the digital realm, in which decision-making is mostly dependent on data is critical to the development of a company. The data has demonstrated its significance in assisting in the development of lucrative business decisions (Silva et al., 2019). The reason for this is that it is based on consumer decisions that may be analysed and affected to make alternative selections. In recent years, several data science use cases in the retail industry have emerged, assisting in uplift and providing a competitive edge in the current trends. Big data often promotes decision-making that is based on the high volume, diversity, and high value of the data acquired by the company (Vermaet al., 2020). Thus, the judgement is mostly based on the quality of the data gathered. Big data analytics researches vast amounts of data to reveal hidden patterns, correlations, and other perceptions of the same. However, the process of data gathering, extraction, processing, and analysis are a very complicated undertaking that frequently needs the assistance of an expert. Therefore, the purpose of this research is to define a big data strategy for the retailer company "Coles," which benefits and improves decision-making and the development of new business models. Furthermore, the research discusses big data business use cases that might be utilised in the retail sector. Lastly, this paper examines the big data-technology and organizational architecture for its solutions extensively.

2 Big Data Use Cases

In the retail business, big data has five significant use cases. Performing consumer behaviour analytics, price optimization, predictive analytics for enhancing conversion rates, recommendation engines, fraud detection, and customer journey analytics are examples of these.

Using Predictive Analysis to Boost Exchange Rates: Predictive analysis is a powerful big data technique that may be used to collect information about clients, such as their hobbies and interactions on social media (Santoro et al., 2018). It is simple to link client profiles, purchase history, and behaviour across social media platforms by applying predictive analytics. The insights gained from this strategy aid in the development and advertising of the retail companies as well as special campaigns on social media pages and TV series.
Analytics of the Customer Experience: Big data engineering solutions incorporate structured and non-structured data into tools named Apache Hive and Hadoop to analyse collections of data regardless of data type. Analytical data is used to increase client retention and drive sales (Félix et al., 2018). Big data insights assist marketers in maintaining constant communications and comprehending the path of each consumer across many marketing channels. The findings of the analysis will reveal new trends in understanding consumer behaviour.
Recommendation Engines: This has proven to be quite useful for retailers because it is a tool that aids in the forecast of client behaviour. Retailers employ recommendation systems as one of the primary effects on their consumers' opinions. Making recommendations will help the store increase sales, retain consumers, and set trends (Dekimpe, 2020).
Detection of Fraud: Data breaches and data frauds are becoming increasingly frequent, posing a serious threat to organisations. As a result, we must identify suspected fraud behaviour using big data technologies and analytics. This information can be gathered through logistic operations and Point-of-Sale transactions. The threat is also related with the private details of the consumers (Carolan, 2018).
Optimizing the Price: As the name implies, having the proper pricing will provide both customers and retailers with a considerable advantage in optimization techniques. The process of pricing development is influenced by the expenses of producing an item, the customer's pocketbook, and the offers of rivals (Aloysius et al., 2018). Data analysis tools are regarded as a new level of approach.

3 Critical Analysis of Big Data Technologies

3.1 Big Data Technologies

There are several big data-technologies that may be employed in retail sector. Below are a few examples:

Hadoop Ecosystem: The Hadoop ecosystem includes all of the technologies that can be utilised to tackle practically all Big Data Analytics concerns. It is an open source software architecture that may be used to store and analyse data on commodity hardware clusters. This system incorporates a number of instruments and features to provide improved record preservation, evaluation, and maintenance. Various procedures in the retail sector may be handled utilising Hadoop in a cost-effective and time-efficient manner. Hadoop is made up of Apache open source projects as well as a number of commercial products (Ying et al., 2021).

Benefits: Hadoop can be used for retail research since it is extremely flexible and proficient of collecting and delivering incredibly huge datasets while also managing several servers running in parallel. Furthermore, the Hadoop Ecosystem includes methods for obtaining significant findings both from unstructured and structured retail data Fahmideh & Beydoun, 2019).
Limitation: Hadoop is not appropriate for handling modest amounts of data. Because it is built for high capacity, it does not handle random reading of tiny files. Furthermore, because many types of data exist in the retail business, the MapReduce framework in Hadoop might have an impact on overall system stability owing to its inefficiencies in dealing with multiple data formats (Silva et al., 2019).

In Memory Databases: Big data analytics solutions can analyse data that is kept in memory instead of on a hard disk drive, which allows it to conduct the process much faster. This is what an in-memory database comprises (Dekimpe, 2020). This will allow Coles to store information and search for insights in a more effective manner.

Benefit: It has speedy queries and real-time upgrades
Limitation: It has administration and stability concerns.

Spark: It is primarily a component of the Hadoop ecosystem that aids in the processing of large amounts of data inside it (Ivanov, & Singhal, 2018). It can assist Coles in identifying shifting consumer behaviour and aids in client retention.

Benefit: It has a very quick processing speed.
Limitation: Spark is too expensive to deploy.

3.2 Data Models

Big data modelling is the act of analysing and building a data model using some formal approach, which encompasses the systematic process of keeping dataset within (Vaganova et al., 2018). The retail organisations need the following data model strategies for its inventory data management:

Logical Data Models: The logical data models include representations of entities, characteristics, relationships, identifiers, subtypes, and super types, as well as relationship restrictions (Gawankar et al., 2020). The logical data model gives a thorough grasp of entities and their interrelationships.

Benefits: In the logical data models, characteristics of each of the specified elements and characteristics of the retail data can be indicated. It enables the use of foreign keys in the overall architecture of databases, allowing entries to be retrieved more efficiently.
Limitation: When constructing the Logical Data Modelling, extreme accuracy is essential because a single error might influence the entire structure.

Physical Data Models: The physical data modelling reflect a broad data architecture since information is transmitted across tables and columns in these models. The retail records are maintained in the usual way, with entities signified by tables and external keys establishing associations between them (Carolan, 2018).

Benefit: The Physical Data Model enables retailers and employees to engage both internally and outside in order to share data connected to retailing records.
Limitation: When it comes to retailing data, the data types might be different for each characteristic; this makes managing the records in this system challenging.