Evolution of Data Analytics
What has happened? — Why did it happen? — What might happen? — What should we do? — What does it mean?
When I first embarked on the digital journey, business intelligence was the revolutionary force. It established the foundation for deriving insights from data, which were then funneled into a decision support system. The system focused on slicing and dicing the historical data, allowing us to measure the performance of key indicators. In turn, this analysis aided in reviewing business decisions made in previous times. However, the landscape has evolved. Now, it is not only about reviewing the past. Today, we expect technology to drive business decisions by processing historical data and constructing probabilistic models that predict and navigate potential futuristic scenarios.
But before we dive into the world of analytics, we should remember that to successfully implement an analytical solution, the emphasis should not solely be on advanced algorithms or state-of-the-art technologies. Instead, the backbone of any potent analytical approach lies in the foundational aspects of data: its collection, integrity, and quality as this impacts the accuracy of analytics.
Having said that, let us look at how it evolved with time.
The Journey started with Descriptive Analytics or “What has happened?”
As the name suggests, Descriptive Analytics looks at historical data to describe past events and current situations. It uses various statistical tools and techniques to explore data and derive insights. This type of analytics serves as the foundational layer in the analytics hierarchy.
For example, by reviewing a company’s past sales history, one might determine that the company’s sales have been on a downward trend. Descriptive analytics often employs data visualization tools to aid understanding. This is the most basic form of analytics which most organizations start with.
To determine ‘Why it has happened?,’ we turn to Diagnostic Analytics.
Diagnostic analytics dives deeper into data to uncover the root causes of specific trends or events. It involves drilling down into datasets, segmenting them, and looking for patterns or anomalies that aren’t immediately obvious. Tools and technologies centered around data mining are commonly used in this analytic approach.
Suppose you are a retail manager and you notice a sudden spike in demand for a particular product. Upon investigation, you discover that the surge in sales is due to a promotional campaign run by the product marketing team. This realization suggests that the increase in demand might not represent a sustained trend but could be a temporary effect that will likely diminish once the campaign ends.
For insights into ‘What might happen?’ in the future, we leverage Predictive Analytics
Predictive Analytics focuses on forecasting future events based on past data.
Now that we have visualized the historical data, identified the trends, and determined the root causes, we aspire to achieve more. We want to look at the future and understand what might happen. This forward-looking approach positions businesses to fulfill customer expectations and do so efficiently.
Various statistical tools and machine learning techniques like regression, classification, etc. are employed to create predictive models. We should remember that no matter how sophisticated the predictive model is, it is based on probabilities and assumptions.
For example, after scrutinizing 24 months of historical demand, we devise a probabilistic model to forecast the subsequent 6 months, thereby effectively planning our manufacturing, procurement, and logistics strategies.
When you need a forward-looking model that provides an optimized action plan and guides ‘What should we do?’, we apply Prescriptive Analytics
Prescriptive Analytics is designed to recommend actionable steps or solutions to handle potential future scenarios.
As we make a gradual transition from a decision-support system to a decision-making system, we increasingly rely on advanced machine learning algorithms like neural nets, deep learning, etc. These technologies assist in navigating multiple scenarios to recommend an optimized solution.
As we unfold Prescriptive Analytics, the concepts of Industry 4.0 are heavily rooted in here.
By integrating with cyber-physical systems, or constructing a Digital Twin businesses can not only digitize their physical operations but also simulate scenarios and generate an optimized plan to strategize their future operations. These in turn help businesses to achieve their goals, enabling more advanced decision-making.
For example, suppose a manufacturing firm is facing issues with inventory management and building large backorders. Prescriptive Analytics can help analyze the data, simulate different scenarios, and recommend the optimized production plan which would help optimize the inventory cost and reduce backorders.
Finally, if you need a human-like reasoning model that answers “what does the data actually mean?” we switch on to Cognitive Analytics.
Cognitive Analytics, powered by Artificial Intelligence and Big Data capabilities, attempts to mimic human-like thought processes in data analysis. It processes large amounts of structured or unstructured data to derive insights and contextual relevance. This is akin to a human expert analyzing a situation, considering past experiences, and making an informed decision.
When combined with technologies like Generative AI and COBOT, it can interact and collaborate with human beings efficiently by generating content and creating a collaborative workspace.
For example, imagine a multinational corporation seeking to expand its operations in a new, unfamiliar market. Cognitive Analytics systems, equipped with Natural Language Processing (NLP), can scan and interpret vast amounts of data from various sources, including news articles, social media feeds, market analyses, etc. By mimicking human thought processes, the system identifies patterns, draws insights, and suggests strategies that would be most aligned with the corporation’s goals and values.
However, it’s essential to understand that AI does not interact with human consciousness but rather processes data in ways that emulate human thinking. To ensure ethical and safe application, Cognitive Analytics should be used in a regulated and governed environment alongside other analytical techniques.
Conclusion:
While the allure of implementing cognitive analytics or prescriptive analytics may be strong, it’s crucial not to bypass the earlier stages of analytics. Each stage corresponds to specific use cases and should be applied as dictated by business needs.
As we advance further into the digital age, the realm of analytics will not only expand but will become a cornerstone in driving impactful and strategic business decisions.
Future businesses will need to integrate real-time analytics, cognitive computing, and advanced technologies to harness the power of analytics to strive for effective business decisions.
References
[1] Duan, ., Da Xu, . Data Analytics in Industry 4.0: A Survey. Inf Syst Front (2021). https://doi.org/10.1007/s10796-021-10190-0 https://rdcu.be/dl1NF
[2] Asakiewicz, Christopher, Cognitive Analytics for Making Better Evidence-Based Decisions (August 24, 2016). Available at SSRN: https://ssrn.com/abstract=2965767 or http://dx.doi.org/10.2139/ssrn.2965767
