Content for the HPC community and innovation enthusiasts: tutorials, news and press releases for users, engineers and administrators
- All News
- Aerospace & Defence
- Artificial Intelligence
- Cloud Platform
- Cloud Platform
- E4 Various
- European Projects
- Kubernetes Cluster
- Latest news
Big Data and Advanced Analytics
In this article we will try to clarify what Big Data are, how to obtain Business Value from them, through Advanced Analytics, and which are the types of Big Data Analytics.
WHAT ARE BIG DATA?
To understand what Big Data are, let’s think about our daily life: interactions with social networks, clicks on websites and our continuously interconnected smartphones generate an extremely high and exponentially growing amount of data. Also, the so-called smart objects, i.e. the new intelligent and connected objects, which produce a constant flow of data detected by on-board sensors: from wearable devices to home automation systems, from Smart City technologies to sensors of Industry 4.0.
The result is that today huge volumes of data are available, extremely heterogeneous in source and format, which can be collected and analyzed in real time: all this gives us a first idea of Big Data.
The scenario is really complex and orienting in this complexity is never trivial, but one fact is now clear for the most competitive companies: knowing how to collect, manage and analyze Big Data is the real challenge of Digital Transformation.
THE 4 “V” OF BIG DATA
Four are the main features of Big Data: VOLUME, VELOCITY, VARIETY AND VERACITY.
Volume is the main feature. To get an idea of the quantity of data involved, just think that there are more than 300 billion photos in Facebook’s systems. IDC predicts that, by 2025, the global data will grow to 163 billion ZettaBytes (1 ZB equals one trillion GB). The growth is very rapid: it is estimated that 90% of the current volume of data has been produced in the last 2 years.
To collect, catalog and analyze Big Data it is often essential to be very fast. As an example, let’s imagine we want to analyze the contents of social channels to plan the marketing of a product; spending too much time in data collection and analysis, would have the unfortunate effect that subsequent marketing campaigns would lag behind the tastes and trends of the moment on the market.
The data to be collected can be extremely heterogeneous: photos, videos, posts on social media, company databases, information from IoT sensors… The variety of Big Data poses non-trivial problems to be solved, both in terms of management and use; complex tools and specific skills are needed to extract truly interesting and significant information from Big Data: Excel sheets are certainly not enough to manage such heterogeneous data!
Although it is the last one, veracity is certainly not the least important. To obtain reliable results from the data analysis, it is necessary that data are reliable. The “Garbage In, Garbage Out” principle applies: if we analyze poor data, so will the results. Given the volume and variety of Big Data, verifying the quality of data is a non-trivial operation and the process can only be automated.
FROM BIG DATA TO VALUE: THE ADVANCED ANALYTICS
ADVANCED ANALYTICS TECHNIQUES ARE THE INDISPENSABLE MAGNIFYING GLASS FOR OBTAINING BUSINESS VALUE FROM DATA, THE FIFTH FEATURE OF BIG DATA.
It is worth starting from the definition of Advanced Analytics given by Gartner: “Advanced Analytics offer the ability to analyze, autonomously or semi-autonomously, data and content, using ‘tools’ that go beyond those of traditional Business Intelligence, with the aim of discovering relationships and correlations, making forecasts, generating recommendations“. This is a definition that highlights some of the key features of Advanced Analytics and, then, underlines how it is an evolution of traditional Business Intelligence (BI).
In general, we can qualify the Advanced Analytics as a decision support system: a good Advanced Analytics solution provides the user with the means to bring to light significant information inherent in the data and transform it into a competitive advantage. An approach to decision based only on intuition and experience is no longer enough today: Advanced Analytics allow you to make decisions based on objective data and help you choose the option with the maximum expected benefit.
ADVANCED ANALYTICS: THE EVOLUTION OF BUSINESS INTELLIGENCE IN THE BIG DATA ERA
BUSINESS INTELLIGENCE HAS BEEN FOR A LONG TIME A VALID SUPPORT FOR BUSINESS; ADVANCED ANALYTICS ARE A CHANGE OF PARADIGM FROM BI AND AMPLIFY THE EXPECTED RESULTS.
BUSINESS INTELLIGENCE (BI)
In traditional BI, business data analysis is built to be repeatable. IT develops reporting models a priori and prepares the procedures necessary to extract the information useful for evaluating the historical performance of the Company: the types of data analyzed and the format in which the information produced is presented are therefore predefined. The result of the analysis is an aggregation of elementary data into “categories” useful for measuring “at a glance” the past performance of the Business. The user of the report then can disaggregate any anomalous result, to identify the causes in detail; traditional analysis starts from the big to get to the small.
With Advanced Analytics, the approach is in fact overturned: the starting point is the user who is looking for answers to a question or wants to clarify a doubt. An Advanced Analytics solution generally provides a friendly interface that allows those without a mathematical and statistical background to work with the data and search for the answers they need. The software will guide the user through the analysis techniques, helping him to select the most appropriate to solve the problem, using, in addition to corporate structured data, also unstructured data, such as social comments, images and videos, which may contain valuable information to “package” a reliable answer.
BIG DATA ADVANCED ANALYTICS TYPES
DESCRIPTIVE ANALYTICS (Study il passato)
Descriptive Analysis uses data aggregation and graphical visualization to provide information on the past and answer the question “what happened?”. Thanks to Descriptive Analytics, the raw data becomes interpretable for the human being, to summarize and describe the different aspects of the Company’s business, analyze past behavior and understand how to improve results in the future.
PREDICTIVE ANALYTICS (Anticipate the future)
Predictive Analysis uses mathematical models and Machine Learning techniques to predict the future and answer the question “what could happen?”. Predictive Analytics combine data to identify patterns and apply mathematical models to explore and capture hidden relationships with the aim of providing an accurate estimate of the probability of a future result.
PRESCRIPTIVE ANALYTICS (Support decisions)
Prescriptive Analysis uses optimization and numerical simulation techniques to formulate useful information to answer the question “what should we do?” Prescriptive Analytics try to quantify the effect of decisions to estimate the results before they are made and recommend the best decision to make.
Deep Learning and Machine Learning processes are applied to the last two types.
ADVANCED ANALYTICS: BUSINESS APPLICATIONS
Advanced Analytics can be used in any sector, in all cases where you want to understand phenomena in detail or improve business processes and strategies. The possible applications are many; in the financial sector, for example, machine learning models can be used profitably to measure a customer’s credit risk or to detect fraud, promptly identifying and blocking any suspicious transactions.
In what other sectors can they be applied? Stay tuned to know more!
In the meantime, to know our solutions for Big Data and Advanced Analytics, please, contact us!