By Anand Mahurkar, Founder and CEO of Findability Sciences
Business executives and investors are unanimous in their belief that Artificial Intelligence (AI) and Machine Learning (ML) will alter their organisations by lowering costs, minimising risks, simplifying processes, driving growth, and sparking innovation.
The potential for AI to drive revenue and profit growth is enormous. The business functions where AI adoption is most common are service operations, product and service development, and marketing and sales, though the most popular use cases span a range of functions as per The State of AI Survey by McKinsey.
Machine learning algorithm scrawl through massive data sets and uncover insights that are hidden to the human eye. By detecting different patterns, businesses transform their raw data into powerful analytics. Understanding the type of data required to make the decisions is central to finding value in the information by organisations. Contrary to the popular belief that relying solely on “big data” for successful deployment of AI has been nullified and stands untrue in today’s world. A shift in focus by organisations towards the concept of adopting “wide data” can change the way AI is used and deployed in organisations drastically.
Why variety of data is valuable to the organisations?
Big data has been used by organisations to focus on analytics that could only tell what happened inorganizations, and hones-in on the so-called “three V’s” — volume, velocity, and variety. That means combining internal, external, structured, and unstructured data. Utilizing a variety of data sources has become mission critical today, where there are many parameters and dependencies beyond an organization’s control. Variety allows organizations to harness the power of multiple data points to obtain meaningful insights, make smarter predictions, and gain valuable analytics for decision making. Variety of data helps ML applications to learn correlations with the factors beyond organization control or beyond the limited set of data which is often used in man-made decisions like observed in the manufacturing sector. AI for manufacturing industry assists in overcoming these draw backs of such traditional approaches.
Big data has made its mark in the field of journalism as well. It not only provided a huge space for data storage and analysis but also made it possible for a convenient process in sourcing of information. With the obtained information, it generates news articles and there by helps in identifying trends. AI for news is without doubt, the next big game in the industry.
Why volume of data is not important?
Not every business or department has loads of data.Plus, Volume is very relative and subjective toevery situation. Modern AI applications can be trained on a relatively small volume of the data. Butfor ML applications to learn, perceive reason and make decisions, more variables are important.Morevariablesmeanmore variety ofdata.
In a connected economy, geopolitical situation, natural disasters, pandemics, socio economic factors can impact the supply chain. If we want to estimate inventories for a manufacturing facility, we need data from varied sources-not only the internal data but all exterior data showing external influences. There by when mixing internal supply chain data with external factors – it culminates into wide-data.
This contextual understanding can be able to predict the reactions or the next move of customers, this unstructured wide data which comes from their social media feed, their repeated behaviors on digital platforms provides actionable insights. This can help enterprises to tap into these insights and offer personalized services.
The use of big data and wide data comes down to how effectively they can be utilized and analyze data scale. After the experience of many failures and successes, it’s learnt that the wide data helps in kick starting effective AI journey much rapidly. Therefore, it changes the dynamics phenomenally for traditional enterprises who may not have sorted out entire data strategy or data platforms. For the musing wide data can be unleash next generation of growth.
Big data, as the name suggests, is big. Its characteristics are volume, frequency, and variety, which lead to its use is somewhat limited to building bigger picture ideas. It’s great for visualizing market trends or understanding the distribution pattern of its components. But it will not go into the details of the trend. This is where small and wide data come in. Small and wide data are better at picking out more specific information and distinct insights from individual data components and drawing valuable comparisons. Wide data provides a variety of small and large, unstructured and structured data. Small data, on the other hand, is focused on applying analytical techniques that bring out useful information within small sets of data. It collects and analyzes data sets sourced within individual organizations or based on individual problem-solving examples.
Large-scale data collection, associated with big data approaches, including collecting vast amounts of data for analytical purposes, is challenging for many organizations. And given the consistent need for data by artificial intelligence, such an approach can be cumbersome for day to day data AI development. In such cases, data analytics need to rely on data sourced recently, specifically and in smaller amounts — that is, small and wide data.
Big data proves very costly for most organisations, is not current, or easily managed. Wide data solves all those three problems. It offers real-time insights, is concise enough to work with speedily, and is not as expensive on the budgets. Additionally, and importantly, it also offers personal insight, which helps the business make more targeted decisions and plans. This is not to say that big data will become obsolete, but more organisations joining the AI bandwagon will steer towards small and wide data.