According to research and consulting company IDC’s “Data Era 2025” report, by 2025 the amount of data in the world will be 10 times greater than in 2015.

Logistics industry trends mirror those of the overall global economy. The amount of logistics information doubles every two years. According to analysts, in 2020 the total amount of data will be 44 zettabytes. For reference, a 1 zettabyte hard drive holds more than 63 million years worth of high resolution 4K video.

To analyze and interpret such large volumes of data, machine learning algorithms come to the rescue. Machine learning helps create forecasts based on available information and trends identified therein. We asked Vitalij Verbilovich, Head of the Research & Development Division at AsstrA, about machine learning and its application in the transport and logistics industry.

Vitaliy, how is machine learning used in logistics?

Machine learning is used in all fields of transport and logistics, including:

• Warehouse Logistics. Computer vision recognizes the presence of goods in warehouses, monitors workers, and provides security at facilities.
• Expedition. Based on the information collected about transportation, plans and routes can be built and seasonal flows can be predicted.
• Sales. Sales volume forecasts can be made taking into account pricing changes from transport and logistics providers and historical sales indicators.
• Security. Scoring models – scoring systems based on statistical methods and supplier relationship information – help identify unscrupulous or potentially problematic contractors even before starting cooperation.

What information is processed using machine learning algorithms? What information should be left for analysts?

In analytics, the first priority is setting a task and formulating a request to select the necessary information. This is not without human intervention – experience and knowledge about a specific industry are necessary. Next, machine learning algorithms come into play. These algorithms cope more efficiently with the collection, processing and primary analysis of information. Analysts are freed from mundane, time-consuming tasks and can focus on the more conceptual aspects of the job.

How does AsstrA use machine learning algorithms?

The AsstrA-Associated Traffic AG corporate group uses machine learning algorithms to solve three types of tasks:

1. Workflow digitalization via the construction of relevant databases with further information processing.
2. Forecasting and flagging possible force majeure on transportation routes. AsstrA partners with Shippeo to increase supply chain transparency; its algorithms enable real-time supply chain transparency and also help predict and warn of possible problems en route.
3. Predictive analysis of indicators from previous periods and assessments of future risks and opportunities.

Thanks to the processed information, better decisions can be made to increase the efficiency of supply chains.