An in Depth Exploration of Machine Learning and Statistical Learning Techniques in Data Analytics

In this article, we describe a data analytics methodology for gleaning insights from the production lines of a power transfer unit, such as the critical measurements needed to construct a shim used to align shafts. This study also outlines the most effective methods and analytical methodologies for domains used in determining which measurements are afflicted by faults; determining which measurements are afflicted by shim dimensions; determining which relationships exist between station codes; forecasting shim dimensions; determining which duplicate samples are present in faulty data; and determining which error distributions are afflicted by measurement. Both statistical analysis and analysis based on machine learning (ML) are used to these domains. These findings demonstrate the reproduction rate of defective units, the relative significance of measurement in relation to the shim dimensions, error distribution and faulty units of measurements. The 'PTU housing measurement' was shown to be the most critical measurement out of all the shim dimensions by both statistical and ML-based analyses