When it comes to the Internet of Things (IoT) it’s all in the data. What really matters though is how you actually go about creating value from your IoT data.

Car manufacturers know what they want from data. They’re wondering if they can understand behavior patterns of drivers, and link that to the wear and tear of parts that need repairs. The key here is not just teasing out the relationship between a driver’s actions and the breakdown of parts, but in the timely alerts that advise drivers to bring their cars to a certain dealer for repairs before a breakdown occurs.

Automakers, and operators of heavy machinery such as those used by oil and gas companies, are increasingly coming to Pivotal with such problems, and the data science team is helping them use Pivotal products and data science to solve them.

The data can be quite varied. For example in drilling for oil, measurements such as pressure on drill bits, rpm’s, etc., are continuously recorded. In automobiles, multidimensional GPS data are used to describe the geo-location of vehicles. As the IoT evolves, the variety of data sources will continue to expand, and applications must step up to be able process data of different kinds, such as sensor, image, and text data. Pivotal Greenplum and Apache HAWQ are uniquely equipped to deal with this such data.

We’ve also implemented complex models and techniques such as Kalman filters for estimating geo-positions, Fourier transforms for frequency decomposition, edge detection for image processing, and window-level aggregations for time-series analyses using open source tools and libraries. Then we leverage MADlib to build models that predict failure. We’ve also built a couple of prototype apps for the consumption of these models which are deployed using Pivotal Cloud Foundry. Thus we are able to provide an end-to-end solution for our IoT customers in oil and gas, automotive and other manufacturing industries.

In the end, the data is all there. For the mechanics of heavy industry to become increasingly effective through real-time machine and sensor data, data science is stepping in as the application response to help untangle the IoT. Data Science can help make systems smart by transforming quality control, improving production line simulation and enabling machines as a service apart from making supply networks smart and improving preventive maintenance. While this changes business processes, the real promise is the new value that can be discovered in using this data in new ways.