Data will have no value if it is not analyzed to extract important information to help businesses (or individuals) increase their operational efficiency. In other words, from raw data to information needs to go through a process – Data Analytics. With the amount of data being generated more and more in both quantity and diversity of data sources and data types, the use of analytical methods, supporting software tools… is increasingly important for every business. In a manufacturing plant, imagine by analyzing errors, causes of downtime, the factory can reduce the number of errors, increase OEE by a few percent, how much has the output increased?

When talking about data analysis, people often talk about the following 5 types:

• Descriptive
• Diagnostic
• Predictive
• Prescriptive
• Cognitive

DESCRIPTIVE ANALYTICS

Descriptive answers the question WHAT has happened in the past (historical data), such as how many shutdowns occurred last month. Either answer the KPIs pass or fail… In short, Descriptive describes what happened (what is here depends on the needs of each company, department). When we have the result, it doesn’t change anything (because it has already happened). At the factory, features like Historical Data, Trend on HMI, SCADA can do these things.

DIAGNOSTICS ANALYTICS

Diagnostic, like the kind of doctor who forces you to do tests before giving a patient a conclusion. Diagnostics answers the question WHY events happened in the past. Diagnostics is an addition to Descriptive, finding the cause of an event will help prevent similar events from happening in the future (this is at the discretion of the business). For example, if Diagnostics answers that machine A fails because of frequent overloads, the operator knows he will need to adjust the load accordingly and will not stop again. Diagnostics also looks for unusual data points, using statistical probabilities techniques to find associations with data trends.

PREDICTIVE ANALYTICS

Predictive means predicting, predicting WHAT WILL happen in the future. It seems more attractive, because Descriptive and Diagnostic only answer about the past, no matter what and why, the incident has already happened, cannot be saved, the damaged product is also broken, the machine stops then stopped. Predictive tells us about the future. Using historical data, look at data trends and compare them with existing real-time data to predict the next trend. Predictive uses many tools of statistical probability, Machine learning… to do these things.

PRESSCRIPTIVE ANALYTICS

This is even more advanced, if you can already predict it, why not make a decision? It is Prescriptive itself (this word in medicine means to prescribe medicine, mean when you find a disease, you can prescribe it, nothing more). Prescriptive answers what needs to be done to achieve the set goal. When people say data-driven decisions, yes, it is. With analytical information, Prescriptive will support faster and more effective decision making for business leaders.

COGNITIVE ANALYTICS

It’s too advanced. Cognitive is awareness. Cognitive analytics combines the power of previous analytics with the context of different scenarios based on a variety of machine learning and AI techniques. Cognitive Analytics simulates how people solve problems. With the power of computers, it is expected to solve many problems better than humans.

Summary, when talking about data analysis – or Data Analytics, we are not only talking in general but need to classify different ‘classes’ not only by technique but also by other resources and different knowledge in this area. The need of Data Analytics is undisputed. The problem is that we need to determine where we are in these 5 types of Analytics and what we want to do more. Define in these 5 categories, the path will be clear. Collecting data without analysis is waste of time. Analysis without clear goals to achieve is also not helpful.