Machine learning algorithms predict properties missing from plastics data sheets

Resin property database provider MobilSpecs LLC has released advanced machine learning algorithms to provide injection molders with predictions for missing data released on plastic material data sheets. The technology was originated from the need for injection molders to have access to more processing information.

"Of the 22,000 materials in our database, 80% have fewer than half of the traditionally reported processing parameters. When injection molders are trying to determine how to set their machines, the lack of reported information forces them to compare against similar materials or guess," said Doug Kenik, managing director of MobileSpecs LLC. "Not only does playing a guessing game cost valuable time, but it also increases risk, so we decided to help processors by leveraging machine learning technology."

The Mobile Specs machine learning algorithms have been able to identify trends in the data to fill in several missing values which are not reported or unknown. The predicted data should provide processors with confidence since the average error for predicted data values is approximately 5%. More conservatively, the data team expects that for 95% of the materials, the true value will be within 15% of the algorithm's predicted value.

The predicted values from MobilSpecs’ machine learning algorithms are not provided by the material suppliers. They are estimates that Mobile Specs has generated in-house with its proprietary Machine Learning technology using publicly available information from the material data sheets.

Processing parameters are estimated based on fundamental properties.


Comments (0)

Please log in or to post comments.
  • Oldest First
  • Newest First
Loading Comments...