A customer want to predict quality of parts to their product(mobile phone) , do we have any case study or customer reference to this use case or similar? And could we use TWX Analytics to meet this requirement?
Anytime you're looking to predict something, TWX Analytics is going to be a tool you might consider using. When you say that the customer is looking to predict the quality of parts, what exactly do mean? Quality can be a subjective term. Some people might consider Kia to be a quality automobile, while others might not.
If you're looking to do predictive analysis around part failure, then TWX Analytics can certainly do this. There is actually a coffee machine demo available from PTC that does this. Kalypso has a demo specific to Wind Farm turbine predictive analytics, where we're able to predict various part failures based on continuously changing data inputs across different objectives. The key is feature engineering--identifying the various data points (manufacturing location, manufacturer, time in service, operating temperature, geographic location, service technician, lot number, etc) that you can use to build a predictive model. Typically the larger the training data set, the better your model and the more accurate your scoring predictions will be. TWX Analytics does a great job of using multiple prediction algorithms and automatically determining the best one to suit your data.
I agree with Wayne's answer. You just need a quantifiable goal (like part failure) in mind and data that has quality features. ThingWorx Analytics can then automatically generate the best predictive model based on the goal that you have in mind.