About the Client
Major manufacturing firm
Case Study
When products coming off an assembly line fail routine tests, manufacturers often employ highly skilled workers to determine the root cause of the failure. These individuals are difficult to train, and therefore are difficult to replace if they leave. As such there is a great opportunity for machine learning models to facilitate and accelerate the discovery of the root causes of failures, easing the burden of discovery on technicians in response to a test failure.
TeraCrunch Solution
Using a large number of historical records detailing product attributes and the root causes of test failures, TeraCrunch developed predictive models for determining root causes, including the specific part that failed and its physical location in the product. Each of the models achieved accuracies above 95% on validation data before they were deployed.
TeraCrunch hosts the solution and provides access to the client through a secure web interface (Tool). In practice, when a product fails a test on the floor, a technician simply provides input to three fields on the web interface tool and receives in return predictions regarding the part that failed and its location in the device.
The predictive models are retrained at regular intervals to capture new products and the failures that are associated with those new products.
Impact On the Business
TeraCrunch tool reduces significant costs as our tool reduces the need to employ highly skilled workers and/or train new workers with this specific skill set.
