In this second post, we cotinue to explain the meaning of other analytic disciplines such as prescriptive analytics or cognitive computing.
PRESCRIPTIVE ANALYTICS
This is the great promise of analytics, although it seems it is never consolidated.
Prescriptive analytics has a big potential to be a disruptive power in business. Along with predictive data, prescriptive analytics can contribute valuable knowledge about what can happen in the future and automate decision-making to achieve business goals. Prescriptive analytics can recommend the products that will lead to the biggest purchases, help choose marketing campaigns that will generate huge profits, determine the optimal routes that generate less costs and minimize the resources used to guarantee delivery times, among others. These solutions can lead to better actions to achieve business goals and enable agile decision-making that makes companies more efficient and competitive. In other words, the suggestions offered by analytics can offer value to a company.
Gartner defines prescriptive analytics as a form of advanced analytics that exploits data and uses complex event processing, a business rules engine and operational research to make better strategic, tactic and operational decisions. While all analytics approaches have been found to improve decision-making, only prescriptive analytics recommend the best solutions.
In essence, the solutions based on prescriptive analytics help to risk limit, increase efficiency and fulfill goals, allowing business owners to make better decisions for their companies even though conditions may be changing or uncertain. They give critical support in operational, tactic and strategic decisions. Prescriptive solutions offer simulation capacities that evaluate the probability of different outcomes and demonstrate which the best option is. Additionally, they provide tools to organizations that help them understand risks better.
For example, Google’s self-driving car works with travel. It makes multiple decisions about what it should do based on predictions of future outcomes. When the car approaches an intersection, it needs to analyze the context of the moment, such as other vehicles, people, positions, routes, distance, the possible routes other drivers will take, etc. The car evaluates every possible option and strives to make the best decision based on the immediate goal: going through the intersection without violating any traffic laws or causing any accidents.
Prescriptive analytics supports three major disciplines:
- Operational research
- Business Rules Management Systems
- Complex event processing
Operational Research
This mathematical discipline was founded by British scientists in World War II. It consists of using different math optimization techniques (lineal programming, mixed-integer programming (MIP), constraint programming, and heuristic algorithms) to solve complex problems, considering all restrictions and limits. The purpose of this mathematics model is to understand the impact of every possible decision. It quantifies each decision in order to choose the best result that will fulfill the target sets.
The optimization models help make complex decisions about cover strategies, supply chain planning, manufacturing schedules, stock optimization, route optimization, campaign optimization and workforce management. In some cases, the models are used to manage resources, reduce costs, increase revenue and maximize profit margins.
Here is another example. Imagine the complexity involved in the maintenance operations planning of a fleet of aircraft. We need to consider the scheduling of every task (with different priorities), stock availability, spare and necessary tools, etc. Many resources are shared and limit accessibility. Additionally, each aircraft has different locations (airports, hangars, runway, routes, etc.). These tasks are carried out by specialized professionals, who are individuals with the necessary skills, time and availability to do the job (always taking in count the regulations and collective bargaining agreements). Furthermore, we should consider the schedules of each aircraft.
Given the competitive prices in the air passenger transport sector, this force minimizes operational costs. This requires that airlines maximize their resources (fleets, materials and humans) while fulfilling legal requirements, guaranteeing flight safety and maintaining flight scheduling. In this case, we can see the complexity and impact of decisions. Any incident can lead to flights being cancelled, a scheduling problem that translates to millions of euros in costs and unhappy customers. This also negatively affects the company’s reputation. The past year, companies like Ryanair and Vueling have had serious trouble with planning, which has had a huge impact on their business. In these cases, math algorithms can help us assess all the data available, consider every restriction and evaluate every option to guarantee the fleet’s schedule.
You can read more about this type of analytics in this link.
Business Rules Management System
Software fields attempts to automatize business logic through decision-making processes in which business rules imply a certain complexity and the result can be known in advance. Business Rules Management Systems (BRMS) are collaborative information systems that centralize the decision-making policies of an organization in only one repository. They permit an easy implementation, some of which can be done in natural language. Additionally, these platforms provide some advantages in the decision process like legibility, transparency and traceability. Thus, these systems are powerful allies in regulated business.
BRMS are used widely in sectors like finance and insurance to make decisions about tariff calculations, subscription processes and risk management, among others.
You can read more about it here: (https://www.decidesoluciones.es/que-es-un-brms-y-sus-ventajas/).
Complex Event Processing
Complex Event Processing is another prescription field where technology provides the capability to analyze data from multiple sources, identify events and patterns, analyze impacts and develop action plans for them in real time. Thus, CEP is a solution that automates decision-making. It is supported in software architecture that handles large amounts of data and variables from different data sources. While CEP was born in the 1990s, it is currently being pushed with initiatives such as Big Data and IoT.
Example “Pay As You Drive”: (https://www.decidesoluciones.es/aplicando-cep-en-el-seguro-para-coches/).
Reach out more information about CEP “Complex Event Processing“: (https://www.decidesoluciones.es/introduccion-al-procesamiento-de-eventos-complejos-i/).
COGNITIVE COMPUTATION
Cognitive computation or cognitive analytics are terms popularized by IBM that combine different analytics techniques within artificial intelligence. But what it is? This field aims to simulate the thinking process and emulate human reasoning through computation modeling. Imitating deductive modelling uses the human brain to extract conclusions and identify patterns through data and knowledge repositories. It also manages the uncertainty of each decision. As I have already indicated, there are some examples where machine learning is included into this cognitive field.
It is necessary to make sense of the data coming from different sources and interpret it before the information is processed. To do this, we help other fields like NLP (Neuro-Linguistic Programming). NLP looks to understand information within a context and interpret natural language the way a human being would.
Theoretically, there are no limits to the cognitive development these systems can achieve, harnessing the benefits of parallel and massive computing.
This technology can be applied in medical care to help collect all existing knowledge about a symptomatology, including patient history, scientific publications, diagnostic tests, etc.
Acting as a virtual assistant supporting possible treatment options, and thanks to this apply the best possible treatment.
In conclusion, this same process could be applied in any field where there are large amounts of data that need to be processed and analyzed to solve problems. Other high-impact business areas include cyber-security, consumer behavior, product recommendation Chatbots, user and customer support bots, travel agents, etc.