Big Data
What Is Big Data?
Working with volumes of data that exceed the computing power or storage of a single computer creates challenges, and big data techniques are used to overcome these. To understand the concept of big data, we must first understand the meaning of “data”.
WHAT IS DATA?
Data in any quantity include characters, symbols, numeric or alphabetic values. Data is necessary for performing any kind of computer operation. It can be transmitted in the form of electric signals and can be stored in optical, mechanical, or magnetic storage devices.
WHAT IS BIG DATA?
As the name implies, “big data” is simply a large amount of data. The main difference between big data and standard data is size. In most cases, big data sets are large and are often growing so rapidly that they cannot be handled with the applications and tools used for handling standard data.
Categories of Big Data
There are three major categories of big data: structured, unstructured, and semi-structured.
Structured data is organized into a relational database with unique identifiers, and it is easy to map and understand. Typically, structured data exists in rows and columns for easy analysis. An example of structured data is sales figures for different products.
Unstructured data doesn't follow any structure or order. Although not ideal for easy analysis, it potentially can be widely used for purposes such as business intelligence. An example of unstructured data is text-based customer feedback.
Semi-structured data does not exist in a database but has relational values and organization that can be analyzed. An example of semi-structured data is text in a document that is marked up or tagged with descriptions, as XML does in a word processing application.
The Three Phases of Big Data
Database management and data warehousing are two primary components of this phase. In phase 1, at the very beginning, big data was stored within a business, often internally within the physical location. With the advent of the Internet, advanced networking allowed for such data centers to become accessible online. The US government created the world's first data center to store 175 million sets of fingerprints and 742 million tax returns.
In phase 2 of big data, web-based unstructured content became the main focus of data analysis for many organizations. The growth of online stores in the early 2000s led to the expansion of web traffic across websites, increasing the amount of data being captured. Companies like Amazon, Yahoo, and eBay started using big data to analyze customer behavior.
The third phase of big data encompasses mobile and sensor-based content. The growth of the smartphone has made it one of the most significant tools for collecting data. Businesses now have the opportunity to analyze the behavior of customers using this data.
Big data can be very beneficial if used in the proper manner. For example, accurate analysis of big data can help a company make the right decisions regarding sales and services. Analysis of customer data in an online store can create a recommendation tool for additional products to the customer. It can also guide price changes for a product or service, based on a customer's budget and spending habits.
If organizations want to use big data effectively to achieve their goals, they must know about its main characteristics. Understanding the “V's” of big data is essential to use it correctly.
The total quantity of data stored is known as volume. The volume of data has grown rapidly. Organizations collect data - structured, unstructured, or semi-structured - from various sources. Some of the sources of data for business are :-
Business transactions.
Outputs from industrial equipment.
Social media platforms.
Smart IoT devices.
Videos.
Data storage was a significant issue for companies in the past, but it is now less of a problem. This is because platforms like Hadoop and Data Lake provide ample storage capacity at an affordable price. These solutions allow a business to store their data in a system that can easily be integrated with analysis tools.
Velocity refers to the speed with which data is created and collected. For example, social networks create vast volumes of data. Facebook users upload hundreds of millions of images every day.
Organizations that want to use this data need processes and systems that can cope with this data, allowing them to derive value from the data without being overwhelmed. Amazon Kinesis is an example of a service that allows organizations to analyze large amounts of streaming data from a variety of sources, including video, website interactions, and IoT devices.
The third V of big data is Variety, which refers to the breadth of sources of data analyzed. For example, the analysis of different types of data related to a customer reveals many things about their needs and preferences.
It is worth noting that variety is not only applicable for customer data but also processes within manufacturing and industry.
A variety of data can help monitor the manufacturing process in steps. If a particular data monitor along an assembly line reports on an issue, it can help pinpoint the source of the problem, rather than requiring the whole process to be troubleshot.
Variability is an additional characteristic of big data. Variability in this context refers to a few different things. One is the number of inconsistencies in the data. They need to be found by anomaly and outlier detection methods for any meaningful analytics to occur.
Big data is also variable because of the numerous data dimensions created from multiple data types and sources. In unstructured data the same word or phrase may mean different things. For example, compare the meaning of the word “great” in these two statements: “I received my order early – great!” and “My order came late and now I see it’s the wrong product – isn’t that great!”
Veracity suggests the quality of the data. Data needs to be consistent, and it needs to be correct. Organizations typically collect data from different sources, which often makes validating and cleansing challenging , but crucial. Outputs from big data analysis will only be as good as the quality of the inputs - “garbage in = garbage out”!
The last V is Value. It is the most important characteristic of big data. The effort taken to gather and analyze big data is only worth it if it helps an organization achieve its goals. In short, the goal of big data must be to add value. For many organizations, this may relate to improving business operations in a range of areas including the customer, the product, or marketing and communication.
Social media, web sites, cloud resources, IoT, and customer relationship management systems (CRMs) are some of the sources of big data that are available for companies.
Big Data Analytics
These various data sources, if they are managed and analyzed correctly, can produce meaningful metrics and information for organizations.
Trends Driving Data Expansion
Online Consumer and Organizational Activity
More people are online, for example using social media. In 2015, there were 2.07 billion active social media users worldwide; that number had increased to 3.81 billion by 2020. With so many users, there is a huge amount of data that is generated every day on social media.
Organizations are increasingly leaning towards using business tools like CRM (Customer Relationship Management) systems that organize and make customer data accessible for analysis.
Smart factories generate large amounts of data to provide automated solutions and maintain self-optimization. The use of such technology can minimize equipment downtime and enhance product inspection.
IoT
Internet of Things is also a growing source of data. Huge amounts of data are generated by devices, such as smartphones, smart home hubs, or smart manufacturing equipment. One of the major advantages of data from the IoT devices is the data is transmitted in real-time.
Big data has been supported by emerging technology and innovations. Let's explore how.
Due to the flexibility of data storage and unconstrained scalability, many organizations now have access to Big Data. Gigabytes of data can be stored and accessed at a much-reduced cost. This facilitates the storage of not only structured but also semi-structured, and unstructured data.
Speed and Accuracy
Additionally, big data streams feed into business processes at an unparalleled speed (velocity), giving on-time and accurate insights. For example, when radio-frequency identification tags (RFIDs) are combined with big data technology, inventory tracking can be enhanced, helping to identify the reason for any delay.
Natural Language Processing (NLP)
NLP, one of the major subsets of artificial intelligence, also aids data scientists. For example, it can be used to process data related consumer and economic sentiment to predict market trends.
Different Applications of NLP
Further, NLP opens the door for studies based on information contained in user reviews, employee evaluations, surveys, news, and other textual sources.
Social Media Data
All our search results, tweets, social media posts are stored on service-providers’ servers. Not only are they stored – but they are also analyzed. This “big data” is used in a variety of ways, for example, to help companies better understand customer behavior.
Big Data Storage a Challenge
Storage and analysis of big data is an important step for businesses, but this can be a challenge. In order to analyze the data, it must be stored, and the storage of big data is a challenging task because big data is enormous in volume.
Let’s Start with Understanding Conventional Big Data Storage Technologies
Distributed File System
A Distributed File System (DFS) is a method of storing and accessing files stored in a server. In DFS, one or more servers save the data, which can be easily accessed by any number of remote clients with proper authorization. A large amount of unstructured data like videos, images, texts, and social media posts can be stored. Hadoop is an example of a distributed file system.
DFS NAMESPACE
Here the administrators can set up multiple DFS Namespaces, where one can map different folder targets in different servers to one particular share path. So, instead of browsing multiple servers to get the files a user needs, just one namespace share can be accessed.
DFS REPLICATION
Here, the user can create a replication group where settings like scheduling and bandwidth throttling that can be configured by the administrator. There is no need to copy files manually.
Cloud Storage
Cloud storage is a relatively new method of storing data. Cloud storage provides flexibility, and both enterprises and end-users can easily use it. For enterprises, the use of cloud storage allows them to control the data from any location at any time, and cloud storage is a cheaper alternative to other storage methods. Similarly, for end-users, cloud storage helps in backups of their data and allows them to use it anywhere anytime.
Things to Know about Big Data Storage
The two most common ways for the storage of big data are HDFS (Hadoop Distributed File System) and Amazon S3.
Apache Hadoop is an open-source software framework that uses a network of many computers to manage and process large amounts of data. Hadoop provides the following benefits:
Helps in speeding up big data processing.
Prevents data loss.
Is a cost-effective method for businesses.
Is highly scalable.
Requires little administration.
Amazon S3, or Amazon Simple Storage Service, has the following benefits:
It is a cloud-based storage service.
It is used by a range of different organizations to store website data, backups, and IoT device data, as well as to carry out data analysis.
It has a range of data security techniques including encryption and provides access to management tools.
Category | HDFS (Hadoop Distributed File System) | Amazon S3 |
|---|---|---|
Storage Type | HDFS uses physical storage, and to increase the storage capacity, more storage needs to be added to the system. The solution becomes more complex as more hard drives are added to the network. It also isn’t very cost-effective as it involves buying new physical storage drives. | In Amazon S3, more storage can be added easily without buying any physical drives. Instead, Amazon charges its consumers every month, making it cost-effective. |
Security | Data on both HDFS and Amazon S3 is secured from unauthorized usage with built-in security features like user authentication and file permissions. | Amazon S3 supports user authentication to control data access. At first, only the bucket and object owners have access to data. Next step permission can be granted to users and groups via bucket policies and Access Control Lists (ACL). This process makes it more flexible than HDFS. |
Performance | In HDFS, data is stored on the same device in which processing is to be done. This usage of local storage makes HDFS faster and more effective. | Amazon S3 uses cloud storage. Therefore, it requires a stable network and compatible devices for smooth performance. |
Shopping Recommendations
For example, how does Amazon provide recommendations while shopping? Most of the time, they recommend things that the user is thinking of buying, or has searched previously. How do they know about their choices? By analyzing big data.
What Is Big Data Analysis?
Big Data analysis includes methods for examining large data sets in order to draw conclusions and predictions that are helpful for organizations. The analysis of big data allows companies to get valuable insights about their customers, competitors, forecasts, and the daily performance of their workforce or machines.
Let’s Explore Some Big Data Analysis Techniques
A/B testing
In A/B testing, a control group is compared with a test group to determine which of two different approaches is more effective. This procedure can improve the marketing response. It is also known as bucket testing or split testing.
Data Fusion and Integration
Data fusion and data integration refer to the analysis of data from multiple sources. Doing so is useful because results derived from various sources are typically better and more accurate than those derived from a single source. For example, companies use data relating to users collected from many sources to build a more complete picture of users, which can then be analyzed.
Data Mining
Data mining is an analysis technique used to extract only useful information from a large set of data, or big data. Using this technique, meaningful trends and patterns are identified and analyzed. Spam filtering, credit risk management, database marketing, analysis of customers’ opinions, and fraud detection all employ data mining.
Machine learning
Machine learning is an aspect of another key technology area, artificial intelligence. Machine learning can be used in data analytics to create predictions based on large data sets, using computer algorithms. These predictions, or models, can learn and become more refined based on new data. Oracle's Machine Learning product family is an example of software that helps businesses to analyze trends in their industries.
NLP is a data analysis technique that uses algorithms to analyze human languages. NLP includes the automatic translation of one language to another, recognition of a spoken word, and answering of questions.
Use of NLP
In the context of big data, NLP is particularly useful for analyzing large volumes of textual information.
Some sectors, for example the legal or medical professions, generate a large amount of diverse documentation, from formal reports to informal notes made by doctors or lawyers. NLP can help to analyze this data.
Statistics
Another approach that supports data analysis is statistics. Statistical techniques that work with smaller sets of data can also be applied to larger data sets, especially now that the computing power to carry out this analysis is increasingly cheap and readily available.
SAMPLING
A sample is taken from the dataset, and model estimates and predictions can be made about the whole dataset based on this sample.
DIVIDE AND CONQUER
The dataset is divided up into small blocks that can be more easily analyzed. The results from the analysis of each of these blocks are combined to produce an analysis for the whole dataset.
Big Data in Decision-making
Big data enhances a company’s decision-making capabilities significantly as it offers ranges of insights that can, for example, improve the quality of service, efficiency, logistics, user preference, forecasting, and profit. However, no matter how insightful the data might be, if the company cannot read it or use it efficiently, then the impact of big data analysis will be reduced.
Data Visualization
For example, non-technical managers who need to make decisions based on data may need tools to help them understand analytics. Data visualization aims to achieve this by presenting data in a graphical way in order to communicate data as well as the relationship between data. Depending on the data you need to present, an appropriate type of visualization is required.
Common Types of Visualization
There are many ways to visually represent data. You should familiarize yourself with different options, recognizing that choosing an appropriate visualization depends on the data being presented, the target audience, and the goal of the visualization. Here are some examples of different ways to visualize data.
Two-dimensional areas are commonly used to show data on a single plane – for example on a map.
Area or Distance Cartograms: Area and distance cartograms use parts of maps such as countries, regions, or states, displaying parameters like demography, populace size, travel times, or other factors: for example, data on the visualization of apparel sales for women in different regions of the country.
If two or more dimensions or variables need to be represented, then multidimensional data visualizations are important. These visualizations are the most common and include pie charts and histograms. However, these simple visualizations might not be suitable, or not may not exploit the potential, of a large data set. Another example of multidimensional data visualization is a scatter plot.
Scatter plot: This model makes use of dots to indicate values for two variables. For example, a scatter plot can be used to show the relationship between age and height for a large group of children. It is also possible to introduce a third variable into a scatter plot by changing the “dots”.
For example, color coding of the dots can be used to show the gender of the child.
This type of visualization method shows how certain data or variables are ranked and how they are interrelated.
Dendrogram: This is a tree-like diagram that shows the hierarchical clustering between sets of data. For example, in medicine, if you want to interpret the clustering between genes or samples, then a dendrogram is useful.
When the focus is on the linkages and relationships between elements data or between data sets, then a network data model is the best choice.
Node-link diagram: These diagrams consist of nodes, or icons, and connections or links between these nodes.
For example, these diagrams can represent social networks that allow like-minded individuals or people with the same interests or goals to be in touch with each other.
These include a start and a finish time and present a descriptive image that shows a variable’s change over time.
Polar area diagram: This diagram is similar to a usual pie chart but it has sectors that have equal angles and these sectors vary from each other depending on how far they stretch from the center of the circle. It is useful for representing cyclical data such as data on days in the week, or months in the year.
Big Data in Healthcare
Big data plays a crucial role in the healthcare sector. Various healthcare organizations have leveraged big data analytics to manage their clinical and physical data for a variety of purposes. An important point to note in this sector is that mishandling of data may have very serious consequences, as the data often pertains to the clinical and personal information of patients. Therefore data security and data protection are crucial.
Structured Medical Data
In this sector, big data often relates to the digitalization and organization of vast quantities of information pertaining to patients’ medical attributes. Since most of the physical and medical attributes can be quantified (like age, height, blood pressure, glucose levels), these are mostly structured data and are used to add value by providing a better medical service.
Let’s go through a few examples to understand the uses better.
Big data is using artificial intelligence for predictive analysis of patients’ treatment. Clinical Decision Support (CDS) software can be deployed to analyze the medical data of a patient. This helps the doctors to prescribe the right drugs and the required medical advice to the patient.
Electronic health records (EHRs), which are digitized medical records, facilitate other applications of big data. EHRs contain patient information like demographics, medical history, allergies, and laboratory test results, which are shared via secure information systems. EHRs can be accessed by any medical practitioner from any part of the world.
In Telemedicine, big data analysis is applied to a large volume of patient data in order to create a virtual clinic, with real-time consulting through video calls. This is particularly effective for cases of low and mild severity, and those who do not need a physical examination. It also helps to provide on-time medical advice in emergencies.
Big Data in Insurance
Insurance is another area where big data is commonly used. The subsectors of insurance that use big data are health, automobile, property and casualty (P&C), and life insurance.
Application of big data in the insurance sector:
RISK ANALYSIS
Car insurance is one area that has been significantly impacted by predictive analysis using big data. Insurance companies use big data to understand the driving style, habits, and the related risks of the driver, thus influencing the insurance premium.
FRAUD DETECTION
Big data helps insurance companies to detect fraud. Systems are now sophisticated enough to detect fraud in real-time, for example by identifying potential fraud before a policy or claim is approved. Data used in the analysis can be drawn from a variety of sources, such as underwriting, claims management, law enforcement, and even other insurance companies. The modeling of fraudulent behavior is complex, as criminals are always looking to exploit systems and processes, but the challenges are increasingly outweighed by the benefits.
Big Data in Manufacturing
In manufacturing, “Industry 4.0” - automation using smart technologies - is a key trend. Big data is a crucial component of this. The use of big data helps manufacturing to:
Maximize production
Reduce costs
Customize production based on demand
The analysis of big data can help to improve productivity. For example, by analyzing data relating to all variables affecting output, production processes can be improved and downtime reduced. By implementing big data in manufacturing, patterns can be found, and problems can be solved. All variables in a production process like cost, efficiency, and responsiveness in a manufacturing unit can be analyzed, and the source of problems can be identified.
Customizing goods and services according to the customer's needs is a challenging task as it requires machine changes and reduced production. To tackle the challenges, companies have started using big data analysis to know the actual needs of the customer. They have shifted from a manufacturing-for-customer model to consumer-to-manufacturing. In this model, the specific requirements of the product’s consumer are anticipated by analyzing data relating to customer trends, requirements, and customer types.
With in-depth analysis, big data can enhance the performance of manufacturing equipment in real-time. Machine downtime can be reduced by setting standards for each device using the insights obtained by big data analysis. In this manner, all kinds of errors caused by machines can be eliminated and production efficiency and quality of products can be improved.
Big Data in Logistics
Smart manufacturing organizations now demand a fast logistics system to deliver their products as soon as possible to their customers. The development of eCommerce platforms has also increased the importance of logistics systems. To improve their functioning, big data can be used. For example, companies like TomTom provide fleet management and logistic software based on big data.
Big data in logistics provide the following benefits:
REAL-TIME TRACKING
With the help of advanced technologies like barcodes, radio-frequency identification tags, and global positioning system devices, collection and analysis of big data can be done in real-time. Logistic companies have the power to track the present location of their vehicles and items.
If the product is not delivered, then both customer and seller can trace it. The implementation of big data has made the delivery process fully transparent and quicker.
WAREHOUSE MANAGEMENT
The unavailability of a specific product on a platform may result in loss of business, and warehouse management is crucial in avoiding this. Big companies like Alibaba, DHL, and Amazon operate their warehouses with the help of analysis based on big data. It helps keep track of every item in real-time and sends instant alerts for out of stock items.
A crucial part of every business is its customers. In order to grow, it should focus on the customer experience. How can businesses understand more about their customers?
Application of Big Data in eCommerce
TARGETED ADVERTISEMENT
Advertising is essential to create brand awareness, but creating ads without research only results in a waste of money. It is necessary to observe various customer-related patterns and trends before creating an advertisement campaign. Targeted and personalized ads increase efficiency and reduce costs. An ad targeted at a potential customer who has proper product knowledge or is genuinely interested will generate more sales. eCommerce companies utilize big data to observe online activity and monitor point-of-sale transactions. This allows the ads to be targeted.
INNOVATION
The analysis of big data helps inform a company about the product and services that have a high chance of being sold on its eCommerce platforms. For example, data on popular items that a significant number of customers buy while visiting a website can be an important input into designing innovative products and services.
The following are some of the areas where the application of big data generates benefits in the public sphere:
Education
Education institutes across the world are also reaping the benefits of big data. Analysis can track students studying online and their behavior pattern, and help identify what promotes and hinder learning. By studying these patterns, education institutes can make decisions that will enhance the productivity of the students.
Fighting Crime
Big data also can assist in the fight against crime. Prosecutions relating to child trafficking, insurance fraud, money laundering, and sexual abuse can be supported by big data analysis.
Environmental Protection
With the help of sensors, researchers can be gathering environmental data with the goal of improving air quality. This data can be used to create a visual representation of pollution, and the information is used by researchers to relate it to geography. This enhances forecasting and helps identify causes.
Here are some key terms that you should be familiar with.
Industry 4.0 – manufacturing using smart technologies
In industry 4.0, companies use big data analysis to analyze huge amounts of data collected from smart sensors connected through cloud computing and IoT platforms. This way, they can identify patterns that help improve efficiency.
Predictive analytics
Predictive analysis uses big data to find meaningful patterns that can predict future events.
Let's Explore Big Data Adoption!
Investment Requirements in Big Data Storage and Analysis
The benefits of using big data for an organization have already been covered. However, there are some challenges like storage and analysis. Organizations are increasingly able to gather large volumes of data, for example, data associated with online activity. In order to use this data, it must be stored.
After collection and storage, the main process of data analysis can begin. The analysis helps in finding hidden trends, but it requires numerous advanced tools, strategies, and competences.
In short, the implementation of big data requires investment. Solutions need to be purchased and experts need to be hired.
Storage and Networking
The first step towards the implementation of big data is the collection and storage of data in a secure location. To deal with large volumes of data, you need a powerful and robust network. Choosing a fast network will help support multitasking.
Hyper-scale Computing Environments
One option is to use storage techniques that are specially designed for big data. For example, big companies like Google, Facebook, Alibaba, Microsoft, and IBM use hyper-scale computing environments with dedicated servers, and attached storage, and processing frameworks like Hadoop. These servers include flash drives to reduce the time taken to store or retrieve data (latency) and improve performance. Alternatively, smaller organizations may opt to use Network Attached Storage (NAS).
Cloud Servers
Data can also be stored on cloud servers. Their use can sometimes impact on latency, but they can be very useful as a backup location for data. Cloud storage can be more expensive than other methods, but it is more flexible as it can be scaled up and down as required.
Processing
Processing is an essential aspect of handling big data. The system used needs to be powerful enough to process big data with ease and in a timely way.
There is a range of software solutions to support processing. For example, Splunk offers cloud processing systems that are best for companies that experience seasonal peaks in data volumes. It is important to choose a solution that meets your requirements in terms of cost, as well as other criteria that may be important, such as the environmental impact of data processing. By 2025, data center energy consumption could use one-fifth of global electricity.
Analytics Software
Analytics software is used to examine data and create useful information based on this analysis. While investing in software, you must keep in mind factors such as security, ease of use, and the type of analysis that you want to carry out.
Descriptive Analytics
In descriptive analytics, large volumes of data are split into reasonable chunks. Techniques like data aggregation and data mining use this data to give a summary of the findings and to show the underlying meaning in the data.
For example, descriptive analysis can be used in targeted marketing, where companies mine historical data of the consumers to analyze the consumer behaviors and engagements with their businesses. MS Excel, MATLAB, and IBM SPSS are some of the tools used here.
Diagnostic Analytics
This analytics method is used to determine why something took place in the past, analyzing the root causes of the event. Though limited, it helps in understanding what factors impacted outcomes and can provide actionable insights.
Predictive Analytics
This is a technique that makes future predictions based on current data. It builds models that work on the descriptive analytics data to predict future outcomes.
Prescriptive Analytics
Prescriptive analysis takes future prediction one step further. Given a specific outcome, it can suggest multiple courses of action to reach that point or it can present all possible outcomes to a particular course of action.
For example, an online taxi service could analyze a customer’s distance to nearby drivers and make suggestions on the quickest option.
Employees
Hiring employees or retaining consultants with the right skill sets is crucial.
Specific skills with tools and languages such as Excel, Power BI, SQL, R, and Python are likely to be highly desirable. Similarly, more specialized competencies relating to artificial intelligence, natural language processing, and the Internet of Things will support the effective implementation of a big data strategy. Developing dedicated staff with an appropriate set of skills and knowledge will prove to be an excellent investment for your organization.
The rapid growth in big data is remarkable. Many businesses are looking for the techniques and solutions that will help in the analysis of extensive structured, semi-structured, and unstructured data.
Unstructured Data Challenge
Big data analysis helps companies to achieve their goals, but it also has challenges. For example, big data can contain both structured and unstructured data, creating difficulties for formal data handling techniques. These challenges must be understood before you start using big data.
What Are the Challenges?
Challenge 1
Data quality is the first challenge in big data analysis. One of the main reasons for the inconsistency in data quality is veracity. Veracity refers to the accuracy and relevance of data. The quality of big data may vary because it is not collected from one source: It may relate to social media, search history, watch history, machine data, tweets, and IoT devices, amongst others. In theory, it is better to collect data from multiple sources, but more sources mean more complications.
In big data analysis, it is tough to achieve consistency because of the velocity, or speed, at which it is processed, and because of its variety.
Velocity is a big challenge because if one is not able to analyze data quickly, then the organization may miss out on opportunities. It will be better to analyze and act on big data quickly before it gets outdated. One option to do this is to use big data software, such as Hadoop, to support timely processing and analysis.
Variety is another significant issue. Big data contains both unstructured and semi-structured data. The study of these kinds of data is more complex and time-consuming than structured data as there is no established relationship in non-structured data; they are not available in an easy-to-access arrangement. Searching for a specific word or sentence in non-structured data will give results with different meanings. For example, even in the same language, the meaning of words depends on local dialect and slang. Managing this variety is important to preserve consistency.
Organizations are, more and more, looking to exploit the potential of their existing sets of data, and the data that they are constantly gathering. However, in order to carry out analysis, they need to bring together this data, which may be stored or managed using different systems or processes. This may create significant obstacles to data analysis.
Big data must be synchronized to use it easily on analytical platforms. What does synchronization mean? The synchronization of data between two locations or devices means that all the updates and changes in the data will update on both the locations or devices. This is important for data consistency and is particularly crucial in large or distributed organizations.
There is a dearth of skilled professionals in the market who have a good understanding of data analysis. Collecting enormous amounts of data from different platforms and buying massive storages to store them is pointless if there is no data scientist or data analyst to exploit the investment. Therefore, it’s extremely important to have the right competences in place as part of a data analytics strategy.
Big data is not just about developing insights - it has the potential to directly generate revenue for companies if used wisely. They can leverage their data to generate income by either selling the data they have directly or by selling tools for analysis.
What makes big data valuable enough to sell as a service?
Behavioral Data Insights
Data collected by observing the behavior patterns of consumers can provide insights into their areas of interest, likes, and dislikes. This data is extremely useful for manufacturers and companies to produce or alter products and services to meet customers’ needs. As long as data privacy regulations are met, companies can monetize this data.
Two Ways to Use This Data
BIG DATA AS A SERVICE
Companies can use big data internally for revenue growth. Facebook is a prominent example. It collects vast amounts of data about its users including age, gender, work to their current location, and email address. Facebook uses this data to generate money, not by directly selling off such detailed information but by placing appropriate ads on the “News Feed” of a particular user, based on their activities and preferences. This user-targeted marketing is one of the major revenues for Facebook. In short, what Facebook is doing is “selling” its big data analysis to its customers by providing targeted advertising opportunities.
INSIGHTS FROM BIG DATA
Companies other than the tech giants may also be able to monetize their data. Insights generated through big data analysis may prove valuable to other organizations. In addition to generating more revenue, supplying big data analytics may help foster new partnerships with compatible companies. For example, financial institution BBVA sells value-added analytics based on anonymized transaction data of its customers to the tourism industry. A point to be noted here is that there is no exchange of ownership of data, and it remains the property of BBVA.
Big data analysis has become easier to carry out, and its benefits are more widely perceived. However, big data analysis raises ethical questions for organizations that are exploiting this data, particularly if it is personal data.
Data that contains private information needs to be protected very carefully to make sure that the data is only used in a legal way. One way to help achieve this is to establish good governance structures around data analysis. Someone or some body within the organization needs to take responsibility for the analysis, and provide oversight of it.
The goals of the analysis should be clearly articulated in advance to avoid scope-creep, and analysis should be focused on this goal. The responsible authority in the organization needs to ensure that measurement and data gathering are in keeping with the goals, and also should ensure that the data is of sufficient quality to achieve the objectives of the analysis.
Analysis is appropriate if it meets the agreed goals, and if these goals are ethically and legally acceptable. In general, great care must be taken regarding information relating to politics, race, or religion. Data relating to this should only be used if there is a clear and justifiable reason.
If used incorrectly, this data may be the source of discrimination. In fact, any personal data – information that can be associated with a specific individual – needs to be treated carefully, and there must be an ethical and legal basis for processing this data. One way to avoid issues relating to personal data is to anonymize it so that it cannot be linked to an individual.
It is important to choose the correct data to feed into analysis – and the correct methods of processing the data. Choices around data and tools should be made on their objective merit and likelihood to produce an accurate and valid outcome – not on their likelihood to produce the desired outcome.
In many jurisdictions, ethical considerations are underpinned by regulatory frameworks. For example, the European Union’s General Data Protection Regulation says, ‘personal data may not be processed unless there is at least one legal basis to do so or unless a data subject has provided informed consent to data processing for one or more purposes.’
So far, this module has covered concepts of big data, considerations for its execution, examples of its implementation, and issues around its adoption. In order to start thinking about how one can apply big data in an organization, the following questions should be asked.
Questions to Consider
There is no point in carrying out complex data analysis just because the data and the technology to do the analysis exists. First, the organization must work on a blueprint that sets out why big data analysis is being implemented, and which areas will benefit. This will help focus and justify investment if it is needed and help to drive a more efficient implementation.
There are many potential sources of data for big data analysis, both inside and outside the organization. Internally, machine data can be acquired from various industrial equipment or sensors in the organization, like a production line. Data on customer behavior online or on social media might be acquired from an external organization.
How data is structured will impact directly on data analysis. Structured data is easier to manage, but there may be more potential and richer insights available in unstructured data.
Cleansing and extraction of data is an important step in big data analysis. Technologies like data mining or natural language processes can contribute to this.
Data can be best presented using engaging infographics that can convey the information effectively. Use applications like Power BI or Tableau to create tables, bar or pie diagrams, histograms, and line graphs. Choose the diagram carefully – the aim is not to create a complicated picture, but to convey the key information clearly.
For example, the chart here shows the distribution of customer satisfaction responses for a tech company:
Big data analytics should not be just a research exercise – it should result in practical improvements that are related to organizational strategy. Also, the analysis may have to be ongoing, to ensure that improvement and enhancement are constantly taking place. In short, the analysis should help the organization do something new, or do something better.
Here are some key terms that you should be familiar with.
GDPR
The EU General Data Protection Regulation on data protection and privacy.
What Is Big Data?
Working with volumes of data that exceed the computing power or storage of a single computer creates challenges, and big data techniques are used to overcome these. To understand the concept of big data, we must first understand the meaning of “data”.
WHAT IS DATA?
Data in any quantity include characters, symbols, numeric or alphabetic values. Data is necessary for performing any kind of computer operation. It can be transmitted in the form of electric signals and can be stored in optical, mechanical, or magnetic storage devices.
WHAT IS BIG DATA?
As the name implies, “big data” is simply a large amount of data. The main difference between big data and standard data is size. In most cases, big data sets are large and are often growing so rapidly that they cannot be handled with the applications and tools used for handling standard data.
Categories of Big Data
There are three major categories of big data: structured, unstructured, and semi-structured.
Structured data is organized into a relational database with unique identifiers, and it is easy to map and understand. Typically, structured data exists in rows and columns for easy analysis. An example of structured data is sales figures for different products.
Unstructured data doesn't follow any structure or order. Although not ideal for easy analysis, it potentially can be widely used for purposes such as business intelligence. An example of unstructured data is text-based customer feedback.
Semi-structured data does not exist in a database but has relational values and organization that can be analyzed. An example of semi-structured data is text in a document that is marked up or tagged with descriptions, as XML does in a word processing application.
The Three Phases of Big Data
Database management and data warehousing are two primary components of this phase. In phase 1, at the very beginning, big data was stored within a business, often internally within the physical location. With the advent of the Internet, advanced networking allowed for such data centers to become accessible online. The US government created the world's first data center to store 175 million sets of fingerprints and 742 million tax returns.
In phase 2 of big data, web-based unstructured content became the main focus of data analysis for many organizations. The growth of online stores in the early 2000s led to the expansion of web traffic across websites, increasing the amount of data being captured. Companies like Amazon, Yahoo, and eBay started using big data to analyze customer behavior.
The third phase of big data encompasses mobile and sensor-based content. The growth of the smartphone has made it one of the most significant tools for collecting data. Businesses now have the opportunity to analyze the behavior of customers using this data.
Big data can be very beneficial if used in the proper manner. For example, accurate analysis of big data can help a company make the right decisions regarding sales and services. Analysis of customer data in an online store can create a recommendation tool for additional products to the customer. It can also guide price changes for a product or service, based on a customer's budget and spending habits.
If organizations want to use big data effectively to achieve their goals, they must know about its main characteristics. Understanding the “V's” of big data is essential to use it correctly.
The total quantity of data stored is known as volume. The volume of data has grown rapidly. Organizations collect data - structured, unstructured, or semi-structured - from various sources. Some of the sources of data for business are :-
Business transactions.
Outputs from industrial equipment.
Social media platforms.
Smart IoT devices.
Videos.
Data storage was a significant issue for companies in the past, but it is now less of a problem. This is because platforms like Hadoop and Data Lake provide ample storage capacity at an affordable price. These solutions allow a business to store their data in a system that can easily be integrated with analysis tools.
Velocity refers to the speed with which data is created and collected. For example, social networks create vast volumes of data. Facebook users upload hundreds of millions of images every day.
Organizations that want to use this data need processes and systems that can cope with this data, allowing them to derive value from the data without being overwhelmed. Amazon Kinesis is an example of a service that allows organizations to analyze large amounts of streaming data from a variety of sources, including video, website interactions, and IoT devices.
The third V of big data is Variety, which refers to the breadth of sources of data analyzed. For example, the analysis of different types of data related to a customer reveals many things about their needs and preferences.
It is worth noting that variety is not only applicable for customer data but also processes within manufacturing and industry.
A variety of data can help monitor the manufacturing process in steps. If a particular data monitor along an assembly line reports on an issue, it can help pinpoint the source of the problem, rather than requiring the whole process to be troubleshot.
Variability is an additional characteristic of big data. Variability in this context refers to a few different things. One is the number of inconsistencies in the data. They need to be found by anomaly and outlier detection methods for any meaningful analytics to occur.
Big data is also variable because of the numerous data dimensions created from multiple data types and sources. In unstructured data the same word or phrase may mean different things. For example, compare the meaning of the word “great” in these two statements: “I received my order early – great!” and “My order came late and now I see it’s the wrong product – isn’t that great!”
Veracity suggests the quality of the data. Data needs to be consistent, and it needs to be correct. Organizations typically collect data from different sources, which often makes validating and cleansing challenging , but crucial. Outputs from big data analysis will only be as good as the quality of the inputs - “garbage in = garbage out”!
The last V is Value. It is the most important characteristic of big data. The effort taken to gather and analyze big data is only worth it if it helps an organization achieve its goals. In short, the goal of big data must be to add value. For many organizations, this may relate to improving business operations in a range of areas including the customer, the product, or marketing and communication.
Social media, web sites, cloud resources, IoT, and customer relationship management systems (CRMs) are some of the sources of big data that are available for companies.
Big Data Analytics
These various data sources, if they are managed and analyzed correctly, can produce meaningful metrics and information for organizations.
Trends Driving Data Expansion
Online Consumer and Organizational Activity
More people are online, for example using social media. In 2015, there were 2.07 billion active social media users worldwide; that number had increased to 3.81 billion by 2020. With so many users, there is a huge amount of data that is generated every day on social media.
Organizations are increasingly leaning towards using business tools like CRM (Customer Relationship Management) systems that organize and make customer data accessible for analysis.
Smart factories generate large amounts of data to provide automated solutions and maintain self-optimization. The use of such technology can minimize equipment downtime and enhance product inspection.
IoT
Internet of Things is also a growing source of data. Huge amounts of data are generated by devices, such as smartphones, smart home hubs, or smart manufacturing equipment. One of the major advantages of data from the IoT devices is the data is transmitted in real-time.
Big data has been supported by emerging technology and innovations. Let's explore how.
Due to the flexibility of data storage and unconstrained scalability, many organizations now have access to Big Data. Gigabytes of data can be stored and accessed at a much-reduced cost. This facilitates the storage of not only structured but also semi-structured, and unstructured data.
Speed and Accuracy
Additionally, big data streams feed into business processes at an unparalleled speed (velocity), giving on-time and accurate insights. For example, when radio-frequency identification tags (RFIDs) are combined with big data technology, inventory tracking can be enhanced, helping to identify the reason for any delay.
Natural Language Processing (NLP)
NLP, one of the major subsets of artificial intelligence, also aids data scientists. For example, it can be used to process data related consumer and economic sentiment to predict market trends.
Different Applications of NLP
Further, NLP opens the door for studies based on information contained in user reviews, employee evaluations, surveys, news, and other textual sources.
Social Media Data
All our search results, tweets, social media posts are stored on service-providers’ servers. Not only are they stored – but they are also analyzed. This “big data” is used in a variety of ways, for example, to help companies better understand customer behavior.
Big Data Storage a Challenge
Storage and analysis of big data is an important step for businesses, but this can be a challenge. In order to analyze the data, it must be stored, and the storage of big data is a challenging task because big data is enormous in volume.
Let’s Start with Understanding Conventional Big Data Storage Technologies
Distributed File System
A Distributed File System (DFS) is a method of storing and accessing files stored in a server. In DFS, one or more servers save the data, which can be easily accessed by any number of remote clients with proper authorization. A large amount of unstructured data like videos, images, texts, and social media posts can be stored. Hadoop is an example of a distributed file system.
DFS NAMESPACE
Here the administrators can set up multiple DFS Namespaces, where one can map different folder targets in different servers to one particular share path. So, instead of browsing multiple servers to get the files a user needs, just one namespace share can be accessed.
DFS REPLICATION
Here, the user can create a replication group where settings like scheduling and bandwidth throttling that can be configured by the administrator. There is no need to copy files manually.
Cloud Storage
Cloud storage is a relatively new method of storing data. Cloud storage provides flexibility, and both enterprises and end-users can easily use it. For enterprises, the use of cloud storage allows them to control the data from any location at any time, and cloud storage is a cheaper alternative to other storage methods. Similarly, for end-users, cloud storage helps in backups of their data and allows them to use it anywhere anytime.
Things to Know about Big Data Storage
The two most common ways for the storage of big data are HDFS (Hadoop Distributed File System) and Amazon S3.
Apache Hadoop is an open-source software framework that uses a network of many computers to manage and process large amounts of data. Hadoop provides the following benefits:
Helps in speeding up big data processing.
Prevents data loss.
Is a cost-effective method for businesses.
Is highly scalable.
Requires little administration.
Amazon S3, or Amazon Simple Storage Service, has the following benefits:
It is a cloud-based storage service.
It is used by a range of different organizations to store website data, backups, and IoT device data, as well as to carry out data analysis.
It has a range of data security techniques including encryption and provides access to management tools.
Category | HDFS (Hadoop Distributed File System) | Amazon S3 |
|---|---|---|
Storage Type | HDFS uses physical storage, and to increase the storage capacity, more storage needs to be added to the system. The solution becomes more complex as more hard drives are added to the network. It also isn’t very cost-effective as it involves buying new physical storage drives. | In Amazon S3, more storage can be added easily without buying any physical drives. Instead, Amazon charges its consumers every month, making it cost-effective. |
Security | Data on both HDFS and Amazon S3 is secured from unauthorized usage with built-in security features like user authentication and file permissions. | Amazon S3 supports user authentication to control data access. At first, only the bucket and object owners have access to data. Next step permission can be granted to users and groups via bucket policies and Access Control Lists (ACL). This process makes it more flexible than HDFS. |
Performance | In HDFS, data is stored on the same device in which processing is to be done. This usage of local storage makes HDFS faster and more effective. | Amazon S3 uses cloud storage. Therefore, it requires a stable network and compatible devices for smooth performance. |
Shopping Recommendations
For example, how does Amazon provide recommendations while shopping? Most of the time, they recommend things that the user is thinking of buying, or has searched previously. How do they know about their choices? By analyzing big data.
What Is Big Data Analysis?
Big Data analysis includes methods for examining large data sets in order to draw conclusions and predictions that are helpful for organizations. The analysis of big data allows companies to get valuable insights about their customers, competitors, forecasts, and the daily performance of their workforce or machines.
Let’s Explore Some Big Data Analysis Techniques
A/B testing
In A/B testing, a control group is compared with a test group to determine which of two different approaches is more effective. This procedure can improve the marketing response. It is also known as bucket testing or split testing.
Data Fusion and Integration
Data fusion and data integration refer to the analysis of data from multiple sources. Doing so is useful because results derived from various sources are typically better and more accurate than those derived from a single source. For example, companies use data relating to users collected from many sources to build a more complete picture of users, which can then be analyzed.
Data Mining
Data mining is an analysis technique used to extract only useful information from a large set of data, or big data. Using this technique, meaningful trends and patterns are identified and analyzed. Spam filtering, credit risk management, database marketing, analysis of customers’ opinions, and fraud detection all employ data mining.
Machine learning
Machine learning is an aspect of another key technology area, artificial intelligence. Machine learning can be used in data analytics to create predictions based on large data sets, using computer algorithms. These predictions, or models, can learn and become more refined based on new data. Oracle's Machine Learning product family is an example of software that helps businesses to analyze trends in their industries.
NLP is a data analysis technique that uses algorithms to analyze human languages. NLP includes the automatic translation of one language to another, recognition of a spoken word, and answering of questions.
Use of NLP
In the context of big data, NLP is particularly useful for analyzing large volumes of textual information.
Some sectors, for example the legal or medical professions, generate a large amount of diverse documentation, from formal reports to informal notes made by doctors or lawyers. NLP can help to analyze this data.
Statistics
Another approach that supports data analysis is statistics. Statistical techniques that work with smaller sets of data can also be applied to larger data sets, especially now that the computing power to carry out this analysis is increasingly cheap and readily available.
SAMPLING
A sample is taken from the dataset, and model estimates and predictions can be made about the whole dataset based on this sample.
DIVIDE AND CONQUER
The dataset is divided up into small blocks that can be more easily analyzed. The results from the analysis of each of these blocks are combined to produce an analysis for the whole dataset.
Big Data in Decision-making
Big data enhances a company’s decision-making capabilities significantly as it offers ranges of insights that can, for example, improve the quality of service, efficiency, logistics, user preference, forecasting, and profit. However, no matter how insightful the data might be, if the company cannot read it or use it efficiently, then the impact of big data analysis will be reduced.
Data Visualization
For example, non-technical managers who need to make decisions based on data may need tools to help them understand analytics. Data visualization aims to achieve this by presenting data in a graphical way in order to communicate data as well as the relationship between data. Depending on the data you need to present, an appropriate type of visualization is required.
Common Types of Visualization
There are many ways to visually represent data. You should familiarize yourself with different options, recognizing that choosing an appropriate visualization depends on the data being presented, the target audience, and the goal of the visualization. Here are some examples of different ways to visualize data.
Two-dimensional areas are commonly used to show data on a single plane – for example on a map.
Area or Distance Cartograms: Area and distance cartograms use parts of maps such as countries, regions, or states, displaying parameters like demography, populace size, travel times, or other factors: for example, data on the visualization of apparel sales for women in different regions of the country.
If two or more dimensions or variables need to be represented, then multidimensional data visualizations are important. These visualizations are the most common and include pie charts and histograms. However, these simple visualizations might not be suitable, or not may not exploit the potential, of a large data set. Another example of multidimensional data visualization is a scatter plot.
Scatter plot: This model makes use of dots to indicate values for two variables. For example, a scatter plot can be used to show the relationship between age and height for a large group of children. It is also possible to introduce a third variable into a scatter plot by changing the “dots”.
For example, color coding of the dots can be used to show the gender of the child.
This type of visualization method shows how certain data or variables are ranked and how they are interrelated.
Dendrogram: This is a tree-like diagram that shows the hierarchical clustering between sets of data. For example, in medicine, if you want to interpret the clustering between genes or samples, then a dendrogram is useful.
When the focus is on the linkages and relationships between elements data or between data sets, then a network data model is the best choice.
Node-link diagram: These diagrams consist of nodes, or icons, and connections or links between these nodes.
For example, these diagrams can represent social networks that allow like-minded individuals or people with the same interests or goals to be in touch with each other.
These include a start and a finish time and present a descriptive image that shows a variable’s change over time.
Polar area diagram: This diagram is similar to a usual pie chart but it has sectors that have equal angles and these sectors vary from each other depending on how far they stretch from the center of the circle. It is useful for representing cyclical data such as data on days in the week, or months in the year.
Big Data in Healthcare
Big data plays a crucial role in the healthcare sector. Various healthcare organizations have leveraged big data analytics to manage their clinical and physical data for a variety of purposes. An important point to note in this sector is that mishandling of data may have very serious consequences, as the data often pertains to the clinical and personal information of patients. Therefore data security and data protection are crucial.
Structured Medical Data
In this sector, big data often relates to the digitalization and organization of vast quantities of information pertaining to patients’ medical attributes. Since most of the physical and medical attributes can be quantified (like age, height, blood pressure, glucose levels), these are mostly structured data and are used to add value by providing a better medical service.
Let’s go through a few examples to understand the uses better.
Big data is using artificial intelligence for predictive analysis of patients’ treatment. Clinical Decision Support (CDS) software can be deployed to analyze the medical data of a patient. This helps the doctors to prescribe the right drugs and the required medical advice to the patient.
Electronic health records (EHRs), which are digitized medical records, facilitate other applications of big data. EHRs contain patient information like demographics, medical history, allergies, and laboratory test results, which are shared via secure information systems. EHRs can be accessed by any medical practitioner from any part of the world.
In Telemedicine, big data analysis is applied to a large volume of patient data in order to create a virtual clinic, with real-time consulting through video calls. This is particularly effective for cases of low and mild severity, and those who do not need a physical examination. It also helps to provide on-time medical advice in emergencies.
Big Data in Insurance
Insurance is another area where big data is commonly used. The subsectors of insurance that use big data are health, automobile, property and casualty (P&C), and life insurance.
Application of big data in the insurance sector:
RISK ANALYSIS
Car insurance is one area that has been significantly impacted by predictive analysis using big data. Insurance companies use big data to understand the driving style, habits, and the related risks of the driver, thus influencing the insurance premium.
FRAUD DETECTION
Big data helps insurance companies to detect fraud. Systems are now sophisticated enough to detect fraud in real-time, for example by identifying potential fraud before a policy or claim is approved. Data used in the analysis can be drawn from a variety of sources, such as underwriting, claims management, law enforcement, and even other insurance companies. The modeling of fraudulent behavior is complex, as criminals are always looking to exploit systems and processes, but the challenges are increasingly outweighed by the benefits.
Big Data in Manufacturing
In manufacturing, “Industry 4.0” - automation using smart technologies - is a key trend. Big data is a crucial component of this. The use of big data helps manufacturing to:
Maximize production
Reduce costs
Customize production based on demand
The analysis of big data can help to improve productivity. For example, by analyzing data relating to all variables affecting output, production processes can be improved and downtime reduced. By implementing big data in manufacturing, patterns can be found, and problems can be solved. All variables in a production process like cost, efficiency, and responsiveness in a manufacturing unit can be analyzed, and the source of problems can be identified.
Customizing goods and services according to the customer's needs is a challenging task as it requires machine changes and reduced production. To tackle the challenges, companies have started using big data analysis to know the actual needs of the customer. They have shifted from a manufacturing-for-customer model to consumer-to-manufacturing. In this model, the specific requirements of the product’s consumer are anticipated by analyzing data relating to customer trends, requirements, and customer types.
With in-depth analysis, big data can enhance the performance of manufacturing equipment in real-time. Machine downtime can be reduced by setting standards for each device using the insights obtained by big data analysis. In this manner, all kinds of errors caused by machines can be eliminated and production efficiency and quality of products can be improved.
Big Data in Logistics
Smart manufacturing organizations now demand a fast logistics system to deliver their products as soon as possible to their customers. The development of eCommerce platforms has also increased the importance of logistics systems. To improve their functioning, big data can be used. For example, companies like TomTom provide fleet management and logistic software based on big data.
Big data in logistics provide the following benefits:
REAL-TIME TRACKING
With the help of advanced technologies like barcodes, radio-frequency identification tags, and global positioning system devices, collection and analysis of big data can be done in real-time. Logistic companies have the power to track the present location of their vehicles and items.
If the product is not delivered, then both customer and seller can trace it. The implementation of big data has made the delivery process fully transparent and quicker.
WAREHOUSE MANAGEMENT
The unavailability of a specific product on a platform may result in loss of business, and warehouse management is crucial in avoiding this. Big companies like Alibaba, DHL, and Amazon operate their warehouses with the help of analysis based on big data. It helps keep track of every item in real-time and sends instant alerts for out of stock items.
A crucial part of every business is its customers. In order to grow, it should focus on the customer experience. How can businesses understand more about their customers?
Application of Big Data in eCommerce
TARGETED ADVERTISEMENT
Advertising is essential to create brand awareness, but creating ads without research only results in a waste of money. It is necessary to observe various customer-related patterns and trends before creating an advertisement campaign. Targeted and personalized ads increase efficiency and reduce costs. An ad targeted at a potential customer who has proper product knowledge or is genuinely interested will generate more sales. eCommerce companies utilize big data to observe online activity and monitor point-of-sale transactions. This allows the ads to be targeted.
INNOVATION
The analysis of big data helps inform a company about the product and services that have a high chance of being sold on its eCommerce platforms. For example, data on popular items that a significant number of customers buy while visiting a website can be an important input into designing innovative products and services.
The following are some of the areas where the application of big data generates benefits in the public sphere:
Education
Education institutes across the world are also reaping the benefits of big data. Analysis can track students studying online and their behavior pattern, and help identify what promotes and hinder learning. By studying these patterns, education institutes can make decisions that will enhance the productivity of the students.
Fighting Crime
Big data also can assist in the fight against crime. Prosecutions relating to child trafficking, insurance fraud, money laundering, and sexual abuse can be supported by big data analysis.
Environmental Protection
With the help of sensors, researchers can be gathering environmental data with the goal of improving air quality. This data can be used to create a visual representation of pollution, and the information is used by researchers to relate it to geography. This enhances forecasting and helps identify causes.
Here are some key terms that you should be familiar with.
Industry 4.0 – manufacturing using smart technologies
In industry 4.0, companies use big data analysis to analyze huge amounts of data collected from smart sensors connected through cloud computing and IoT platforms. This way, they can identify patterns that help improve efficiency.
Predictive analytics
Predictive analysis uses big data to find meaningful patterns that can predict future events.
Let's Explore Big Data Adoption!
Investment Requirements in Big Data Storage and Analysis
The benefits of using big data for an organization have already been covered. However, there are some challenges like storage and analysis. Organizations are increasingly able to gather large volumes of data, for example, data associated with online activity. In order to use this data, it must be stored.
After collection and storage, the main process of data analysis can begin. The analysis helps in finding hidden trends, but it requires numerous advanced tools, strategies, and competences.
In short, the implementation of big data requires investment. Solutions need to be purchased and experts need to be hired.
Storage and Networking
The first step towards the implementation of big data is the collection and storage of data in a secure location. To deal with large volumes of data, you need a powerful and robust network. Choosing a fast network will help support multitasking.
Hyper-scale Computing Environments
One option is to use storage techniques that are specially designed for big data. For example, big companies like Google, Facebook, Alibaba, Microsoft, and IBM use hyper-scale computing environments with dedicated servers, and attached storage, and processing frameworks like Hadoop. These servers include flash drives to reduce the time taken to store or retrieve data (latency) and improve performance. Alternatively, smaller organizations may opt to use Network Attached Storage (NAS).
Cloud Servers
Data can also be stored on cloud servers. Their use can sometimes impact on latency, but they can be very useful as a backup location for data. Cloud storage can be more expensive than other methods, but it is more flexible as it can be scaled up and down as required.
Processing
Processing is an essential aspect of handling big data. The system used needs to be powerful enough to process big data with ease and in a timely way.
There is a range of software solutions to support processing. For example, Splunk offers cloud processing systems that are best for companies that experience seasonal peaks in data volumes. It is important to choose a solution that meets your requirements in terms of cost, as well as other criteria that may be important, such as the environmental impact of data processing. By 2025, data center energy consumption could use one-fifth of global electricity.
Analytics Software
Analytics software is used to examine data and create useful information based on this analysis. While investing in software, you must keep in mind factors such as security, ease of use, and the type of analysis that you want to carry out.
Descriptive Analytics
In descriptive analytics, large volumes of data are split into reasonable chunks. Techniques like data aggregation and data mining use this data to give a summary of the findings and to show the underlying meaning in the data.
For example, descriptive analysis can be used in targeted marketing, where companies mine historical data of the consumers to analyze the consumer behaviors and engagements with their businesses. MS Excel, MATLAB, and IBM SPSS are some of the tools used here.
Diagnostic Analytics
This analytics method is used to determine why something took place in the past, analyzing the root causes of the event. Though limited, it helps in understanding what factors impacted outcomes and can provide actionable insights.
Predictive Analytics
This is a technique that makes future predictions based on current data. It builds models that work on the descriptive analytics data to predict future outcomes.
Prescriptive Analytics
Prescriptive analysis takes future prediction one step further. Given a specific outcome, it can suggest multiple courses of action to reach that point or it can present all possible outcomes to a particular course of action.
For example, an online taxi service could analyze a customer’s distance to nearby drivers and make suggestions on the quickest option.
Employees
Hiring employees or retaining consultants with the right skill sets is crucial.
Specific skills with tools and languages such as Excel, Power BI, SQL, R, and Python are likely to be highly desirable. Similarly, more specialized competencies relating to artificial intelligence, natural language processing, and the Internet of Things will support the effective implementation of a big data strategy. Developing dedicated staff with an appropriate set of skills and knowledge will prove to be an excellent investment for your organization.
The rapid growth in big data is remarkable. Many businesses are looking for the techniques and solutions that will help in the analysis of extensive structured, semi-structured, and unstructured data.
Unstructured Data Challenge
Big data analysis helps companies to achieve their goals, but it also has challenges. For example, big data can contain both structured and unstructured data, creating difficulties for formal data handling techniques. These challenges must be understood before you start using big data.
What Are the Challenges?
Challenge 1
Data quality is the first challenge in big data analysis. One of the main reasons for the inconsistency in data quality is veracity. Veracity refers to the accuracy and relevance of data. The quality of big data may vary because it is not collected from one source: It may relate to social media, search history, watch history, machine data, tweets, and IoT devices, amongst others. In theory, it is better to collect data from multiple sources, but more sources mean more complications.
In big data analysis, it is tough to achieve consistency because of the velocity, or speed, at which it is processed, and because of its variety.
Velocity is a big challenge because if one is not able to analyze data quickly, then the organization may miss out on opportunities. It will be better to analyze and act on big data quickly before it gets outdated. One option to do this is to use big data software, such as Hadoop, to support timely processing and analysis.
Variety is another significant issue. Big data contains both unstructured and semi-structured data. The study of these kinds of data is more complex and time-consuming than structured data as there is no established relationship in non-structured data; they are not available in an easy-to-access arrangement. Searching for a specific word or sentence in non-structured data will give results with different meanings. For example, even in the same language, the meaning of words depends on local dialect and slang. Managing this variety is important to preserve consistency.
Organizations are, more and more, looking to exploit the potential of their existing sets of data, and the data that they are constantly gathering. However, in order to carry out analysis, they need to bring together this data, which may be stored or managed using different systems or processes. This may create significant obstacles to data analysis.
Big data must be synchronized to use it easily on analytical platforms. What does synchronization mean? The synchronization of data between two locations or devices means that all the updates and changes in the data will update on both the locations or devices. This is important for data consistency and is particularly crucial in large or distributed organizations.
There is a dearth of skilled professionals in the market who have a good understanding of data analysis. Collecting enormous amounts of data from different platforms and buying massive storages to store them is pointless if there is no data scientist or data analyst to exploit the investment. Therefore, it’s extremely important to have the right competences in place as part of a data analytics strategy.
Big data is not just about developing insights - it has the potential to directly generate revenue for companies if used wisely. They can leverage their data to generate income by either selling the data they have directly or by selling tools for analysis.
What makes big data valuable enough to sell as a service?
Behavioral Data Insights
Data collected by observing the behavior patterns of consumers can provide insights into their areas of interest, likes, and dislikes. This data is extremely useful for manufacturers and companies to produce or alter products and services to meet customers’ needs. As long as data privacy regulations are met, companies can monetize this data.
Two Ways to Use This Data
BIG DATA AS A SERVICE
Companies can use big data internally for revenue growth. Facebook is a prominent example. It collects vast amounts of data about its users including age, gender, work to their current location, and email address. Facebook uses this data to generate money, not by directly selling off such detailed information but by placing appropriate ads on the “News Feed” of a particular user, based on their activities and preferences. This user-targeted marketing is one of the major revenues for Facebook. In short, what Facebook is doing is “selling” its big data analysis to its customers by providing targeted advertising opportunities.
INSIGHTS FROM BIG DATA
Companies other than the tech giants may also be able to monetize their data. Insights generated through big data analysis may prove valuable to other organizations. In addition to generating more revenue, supplying big data analytics may help foster new partnerships with compatible companies. For example, financial institution BBVA sells value-added analytics based on anonymized transaction data of its customers to the tourism industry. A point to be noted here is that there is no exchange of ownership of data, and it remains the property of BBVA.
Big data analysis has become easier to carry out, and its benefits are more widely perceived. However, big data analysis raises ethical questions for organizations that are exploiting this data, particularly if it is personal data.
Data that contains private information needs to be protected very carefully to make sure that the data is only used in a legal way. One way to help achieve this is to establish good governance structures around data analysis. Someone or some body within the organization needs to take responsibility for the analysis, and provide oversight of it.
The goals of the analysis should be clearly articulated in advance to avoid scope-creep, and analysis should be focused on this goal. The responsible authority in the organization needs to ensure that measurement and data gathering are in keeping with the goals, and also should ensure that the data is of sufficient quality to achieve the objectives of the analysis.
Analysis is appropriate if it meets the agreed goals, and if these goals are ethically and legally acceptable. In general, great care must be taken regarding information relating to politics, race, or religion. Data relating to this should only be used if there is a clear and justifiable reason.
If used incorrectly, this data may be the source of discrimination. In fact, any personal data – information that can be associated with a specific individual – needs to be treated carefully, and there must be an ethical and legal basis for processing this data. One way to avoid issues relating to personal data is to anonymize it so that it cannot be linked to an individual.
It is important to choose the correct data to feed into analysis – and the correct methods of processing the data. Choices around data and tools should be made on their objective merit and likelihood to produce an accurate and valid outcome – not on their likelihood to produce the desired outcome.
In many jurisdictions, ethical considerations are underpinned by regulatory frameworks. For example, the European Union’s General Data Protection Regulation says, ‘personal data may not be processed unless there is at least one legal basis to do so or unless a data subject has provided informed consent to data processing for one or more purposes.’
So far, this module has covered concepts of big data, considerations for its execution, examples of its implementation, and issues around its adoption. In order to start thinking about how one can apply big data in an organization, the following questions should be asked.
Questions to Consider
There is no point in carrying out complex data analysis just because the data and the technology to do the analysis exists. First, the organization must work on a blueprint that sets out why big data analysis is being implemented, and which areas will benefit. This will help focus and justify investment if it is needed and help to drive a more efficient implementation.
There are many potential sources of data for big data analysis, both inside and outside the organization. Internally, machine data can be acquired from various industrial equipment or sensors in the organization, like a production line. Data on customer behavior online or on social media might be acquired from an external organization.
How data is structured will impact directly on data analysis. Structured data is easier to manage, but there may be more potential and richer insights available in unstructured data.
Cleansing and extraction of data is an important step in big data analysis. Technologies like data mining or natural language processes can contribute to this.
Data can be best presented using engaging infographics that can convey the information effectively. Use applications like Power BI or Tableau to create tables, bar or pie diagrams, histograms, and line graphs. Choose the diagram carefully – the aim is not to create a complicated picture, but to convey the key information clearly.
For example, the chart here shows the distribution of customer satisfaction responses for a tech company:
Big data analytics should not be just a research exercise – it should result in practical improvements that are related to organizational strategy. Also, the analysis may have to be ongoing, to ensure that improvement and enhancement are constantly taking place. In short, the analysis should help the organization do something new, or do something better.
Here are some key terms that you should be familiar with.
GDPR
The EU General Data Protection Regulation on data protection and privacy.
