Introduction to PMML and its powerful features

PMML Introduction

If someone asks you if you are using a predictive analysis, you may be able to answer "no". Not really, you may be using predictive analytics every day, but you don't know anything about it. When you swipe a credit card or use a credit card online, a predictive analysis model checks whether the transaction is fraudulent. If you're renting DVDs online, it's probably a predictive analytics model that recommends a special movie for you. In fact, predictive analysis has become a part of our lives, and its application will certainly provide you with more help in the future.

With the generation of bridges, buildings, industrial processes and mechanical sensing data, predictive solutions can provide a more secure environment in which forecasts can warn you before potential failures and problems occur. Sensors can also be used to monitor humans, for example in intensive care wards. IBM and University of Ontario Institute of Technology are now collaborating to implement a data analysis and predictive solution for monitoring preterm infants, in which biomedical reading can detect critical life infections by up to 24 hours in advance.

But can predictive analysis only work? Depending on the situation. Open standards are one of the most important components. To enable you to fully enjoy the benefits of predictive solutions and data analysis, systems and applications need to easily exchange information through the following standards. PMML supports the sharing of predictive analysis models between applications and systems.

The main analysis of suppliers ' adoption of PMML is a typical example in supporting interoperability companies. IBM, SAS, MicroStrategy, Equifax, NASA, and Zementis are all members of the Data mining Group (Database Mining group,dmg), and DMG is the Commission that makes the PMML. Open source companies such as Knime and Rapid-iare are also members of the Commission. PMML can shape the world of predictive analysis and make it a better place for you.

PMML Basic Knowledge

PMML is a fact-standard language used to render data mining models. Predictive analysis models and data mining models refer to the terminology of mathematical models that use statistical techniques to understand the hidden patterns in a large number of historical data. The predictive analysis model uses the knowledge acquired in the stereotypes process to predict whether there are known patterns in the new data. PMML allows you to easily share the predictive analysis model between different applications. Therefore, you can stereotype a model in a system, express it in PMML, and then move it to another system, and use the above model to predict the possibility of machine failure in the system.

PMML is the product of a data mining group, a vendor-led committee composed of various commercial and open source analysis companies. As a result, most of today's leading data mining tools can be exported or imported to PMML. As a mature standard that has developed for more than 10 years, PMML can present both statistical techniques (such as artificial neural networks and decision trees) for understanding models from the data, as well as preprocessing of raw input data and reprocessing of model outputs (see Figure 1).

Figure 1. PMML contains data preprocessing and data reprocessing, and the predictive model itself