Industrial Evolution started with the phenomenon like IIOT 4.0 & Maintenance 4.0 back in 2011. This was proposed to integrate, automate and develop predictive model to analyze real world sensory data and report any issues in health of industrial equipment. Out of many public datasets available to determine bearing fault, CWRU is well known and systematic dataset which have been used to train various ensemble learning based models. Various faults like inner race, outer race, ball fault have been analyzed through plotting. Vibration sensor readings were compressed to reduce the size of dataset. Around 20 features have been extracted from CWRU dataset, out of that 11 features were most crucial which are used to train & test various ensemble learning based models to determine type of fault. Most of ensemble based models provided accuracy more than 90%, also Regression based approach was used to determine size of fault which achieved more than 90% of \(R^2\) value.