Is it possible to predict defective products before the process or determine which production phases cause defective products, by examining the production process and defective products?
Purpose: Goal of this project is to reveal the features of manufacturing that leads to poor quality. Various sensor datum and new produced features are examined with different models to predict defective products. Models and analysis results are examined according to expected value framework.
EV represents Expected Value and Expected Value equals to difference between Expected Benefit minus Expected Loss
So according to this formula the gap between the total cost of the defective product before the project and after the project must be higher than the cost of the project in order to be profitable. Also models used to analyse or predict outcomes should not have high false positive rates.
This repository is created for a bootcamp practical case given by patika.dev and Vitra. The dataset used in this project is provided by Vitra.