Automatically Filter Incoming Customer Data with Text Classification
The National Highway Traffic Safety Administration (NHTSA) fined Fiat-Chrysler $175 Million in 2015 for failing to comply with the TREAD Act.
Under the TREAD Act, automotive consumer complaints must be reported to NHTSA in accordance with specific guidelines. Complaints must be filtered and registered according to topic and level of severity.
Automotive Companies gather these complaints in multiple ways, including emails, comments, forms, etc. They must then read and process the complaints to identify potential safety issues and to generally comply with national standards. Manual reporting consumes resources and slows response.
The Lymba Solution
Implemented deep learning techniques with semantic features relevant to the domain. Our client used our pipeline to identify key aspects of complaints submitted, which were then reported to the National Highway Traffic Safety Administration.
Key Features
Automatically generate a report listing model, year, VIN, complaint validity and complaint NHTSA code to feed into company’s regulatory reporting system.
Outcome
Text classification of over 90% accuracy.