An overview of predictive analytics in industrial production
The growing use of social networks, smartphones that collect and continuously generate data, the growing use of the Internet, the presence of sensors that measure and monitor everything, causes the volume of the produced data is growing exponentially, providing valuable information for society and for companies.
All this is Big Data, defined as a large collection of data volume and variety can not be managed with traditional database management tools, but require the use of new technologies and adequate data management systems for storing and analysis, are able to extract their value quickly.
With Big Data are experiencing a new revolution, the large amount of data and information available to us, are considered “black gold” of the new millennium.
They are fundamental to the predictive analysis and extrapolation of information (Data Mining) developed by research institutes and companies in support of their decision-making strategies.
In business intelligence is changing the way to manage information for decision support, they are developing new tools, and down the costs of data collection systems, storage and processing.
As mentioned Big Data are having great success in the analysis Predictive (Predictive Analytics), which is a variety of techniques that predict future results based on the analysis of historical and current data .
This type of analysis, using predictive techniques on machine learning models, is taking place in different sectors; for example in industrial production, marketing, in finance and in the energy sector.
An approach to Big Data in industrial production
The great potential offered by Big Data and Predictive Analytics, make it possible to resolve some issues by large corporations that ultimately are investing capital and resources in this direction, hiring data scientists, funding research projects and enhancing their data infrastructure, to predict failures and anomalies in the operation of certain machines based on the flow of data recorded by the sensors and optimize production.
With the Internet of Things (IoT) and technologies for the analysis and management of Big Data, such as Hadoop and Spark, companies can capture and process an amount of data increasing and this implies the need to identify models for predictive maintenance can improve decision support strategies..
This implies the development and improvement of machine learning algorithms used in Predictive Analytics, such as the supervised algorithms, which develop of classification rules for training using a set of data known (labeled data) or unsupervised algorithms that try to locate the information contained within the data not previously known.
The main type of supervised techniques include neural networks, decision trees, Bayesian classifier or Support Vector Machine (SVM) but also linear and nonlinear regression algorithms. While unsupervised learning is based on cluster analysis using techniques such as K-Means clustering.
To process daily huge data moles it is necessary that the chosen models are very precise, and you have to create prediction models tailored to the type of manufacturing process to be controlled, to promote the reliability and performance. In the next image are shown stages of Predictive Analytics.
Many multinational or major companies are investing in these technologies bringing great profits, or provide services to maximize the potential of Big Data and predictive analysis.
For example, Amazon offers services to use machine learning technology and by AWS provides cloud computing services that help create, protect and distribute applications for Big Data and using these technologies, by recommendation systems, can understand the interests of customers obtaining huge profits.
In the transport sector Trenitalia has announced that it has invested in collaboration with SAP in an innovative project, the Internet of Things and the application of Predictive Analysis software on Big Data, called “Dynamic Maintenance Management”.
The system is based on hundreds of micro-sensors placed inside the trains, which provide information about board components by capturing a large volume of data, which is then reworked with machine learning models applied to maintenance.
With this investment will provide a significant improvement in service and reduce maintenance costs of traditional .
Doing predictive maintenance with Big Data, help avoid breakdowns and production blocks which could cause major economic losses for businesses but also to program the various production stages, making them less expensive and more efficient.
Furthermore, a big advantage that you could get, would reduce, if not eliminate, the time and cost of testing processes, having provided via the Predictive Analytics, the quality of the products during the production process .
The advent of Big Data and Machine Learning techniques is the technological evolution that will change in the coming years, the methods and strategies of the classical industrial production systems.
The versatility of these technologies enables applications in various fields such as in medicine, which is already used for diagnosing and predicting the spread of epidemics, or in the management of energy consumption within the Smart Grid, they are already popular in the marketing and in advertising with recommendation systems able to profile customer preferences, as well in sports, especially team sports through the analysis of historical information, the play action manage to extract valuable information and improve team strategies.
Bibliography and sources of inspiration for this work
-  Beyond the hype: Big data concepts, methods, and analytics. Amir Gandomi, Murtaza Haider. Ted Rogers School of Management, Ryerson University, Toronto, Ontario M5B 2K3, Canada.
-  Predictive Analytics, the power to predict who will clock, buy, lie, or die. Eric Siegel.
-  Using Big Data for Machine Learning Analytics in Manufactoring. TATA Consultancy Services.
-  www.digital4.biz/executive/business-case/trenitalia-investe-nella-manutenzione-che-anticipa-i-guasti.