Machine learning is one of the break-through technologies of the modern digital world. It's applications are found in various research domain such as medicine, image processing, production and manufacturing, aviation and autonomics and many more. To efficiently run a machine, it's maintenance and its monitoring automation system play a key role. The major problem we are targetting is to overcome the lack of an automation system which can give an accuracy rate of the production machine at a given instance of time. Also, the important energy meter parameters required to make power report in an automation system for addressing the production issues, at a given interval of time, were also not recorded. Thus in this paper, we describe how machine learning techniques are used for prediction of the accuracy of running production machine. To address these issues, we have used supervised machine learning technique of Binary decision tree using CART method and for power report, while the data is fetched using RS232 to RS485 convertor via Modbus communication protocol. Using CART we have predicted the machine accuracy at a given time with specific energy meter readings as its input features. This paper discusses the problem definition identified, data analysis of energy meter data and it's fetching and at the end ML techniques applied to predict the accuracy of running production machine. In the end, we prepare various power reports of the different machines from the fetched parameters as well as produce a graphical warning of deteriorating performance of the machine at a given instance of the time.