- 90% of car bodies that pass painting booth have one of 13 paint defects
- Some defects need to be heavily reworked – to the extent of re-painting the body.
- Temperature and Relative Humidity settings in the booth are sub-optimal
- Grant power to the user : A platform that uses ML powered models to predict outcomes, simulate scenarios, take informed actions.
- Prevention is better than cure : Identify optimal operating control ranges so that the plant operator can prevent a potentially defective body from entering the booth.
Highlights of our work
Business ‘wowed’ by
The exploratory data analysis revealed findings that surprised veterans who worked 20+ years at the plant
Rigorous machine learning
Our rigor leaves no stone unturned. We built 22 ML models, and each model was tested using 10+ algorithms
customized to user
We design the user interfaces by putting ourselves in the feet of the user