The power of combining data science tools applied to Internet of Things in freight

Share this post:

How data science enabled a freight company to get its containers back in order.

Data Science is exploratory by nature. You therefore want access to the widest possible range of algorithms, libraries and flexible tools that can handle multiple types of data. But also you do not want to forget that much value can come from seemingly humble techniques such as data visualization that allow a person to spot interesting patterns faster than any algorithm.

For example: we were trying to understand why measurements of sensors on board containers where not arriving on time in a central early warning system. This involves a complex chain of systems and events, stretching from the sensors, via transmitter, satellite, receiver and several IT systems to a central database. After much cleaning and integration of different (types of) data sources we used unsupervised as well as supervised learning techniques to inform us on the major bottlenecks and possible root causes. This effort was successful and actually disproved a long-held believe of a cause of the issue. However, one other, very valuable finding from this project was only indirectly related to the direct project objective. It was a finding that made (again) clear to me what the power of data visualization can be. While building some geospatial visualization of the sensor locations our naked eye noticed something that no algorithm had. We noticed that different sensors, supposedly on the same ship, were moving on different courses across the ocean…! Clear as day light this pointed to a major data quality issue in the administration of sensor to container mapping.

The challenge for IBM Data Science Experience

To resolve your biggest challenges in terms of data, IBM has put together the latest tools and advanced expertise in a collaborative environment developed by data scientists. IBM Data Science Experience is a cloud environment offering everything a data scientist needs for success. It gives you access to a managed Spark cluster, multiple data sets and a combination of open source analytics packages and languages including: Scala/Python/R/SQL, Jupyter Notebooks, RStudio IDE, Shiny and H2O (machine learning).

The benefits

– You benefit from products, tutorials and expertise to improve your skills and knowledge;
– You create value more quickly by exploiting the best Open Source software and innovations designed for data scientists;
– You have a fast path to deployment through IBM Machine Learning;
– You improve productivity thanks to collaborative tools and networking within teams and communities.

“Making data science successful requires an open, engaging environment that encourages collaboration. This is the very essence of what IBM is offering.

You can master the art of data science too, adopting tools created by and for data scientists, with IBM Data Science Experience.