Machine Learning Driven Analytics

Machine Learning Driven Analytics


  • Monitoring and changing the task statuses in databases for started and completed tasks by equipment location geo-fence data
  • Identification and processing of the performing equipment
  • Idling / Operating / Driving time calculation
  • ETA for task completion calculation
  • Identification of a position for equipment engagement / disengagement based on GPS data
  • Processing and storing real-time data to an external database for live monitoring on user's mobile device.


Monitoring. To monitor which part of the field the device has processed, the field is divided into a multipolygon. The tractor is also a polygon, after which the polygon of the tractor is subtracted from the field, and a new shape of the field is calculated.

Real-time data collection. We used Spark streaming to implement real-time data collection from the equipment and devices using GPS.

Testing. To test the algorithm, we have been generating multi-polygons routes with coordinates and sent this data to the server.

Technology stack

Postgres, Postgis, Scala, Apache Spark, Apache Kafka, Nifi, some MAGIC Java and Scala Libraries.


The application tracks interactions between farming equipment in real-time, activities of farming equipment on locations and fields was successfully developed. Among the capabilities of the present version:

  1. Tracking the equipment interaction
  2. Tracking the equipment entering / leaving locations/fields
  3. Calculation of activities / cycles / tasks.