Manufacturing Quality Improvement with Data Mining Outlier Approach against Conventional Quality Measurements

Background/Objectives: This research study has been conducted in a scientific way to help manufacturing engineers and the management team to find out the hidden information from the data which are generated during the everyday manufacturing process. Methods/Statistical analysis: The methodology adopted in this activity is applying outlier analysis which is a data mining technique. The inter quartile range findings and analysis has been used here to find the hidden useful information from the process data with which a better insight could be established towards the improvement of quality of the product. The data used here have been collected from the automotive engine assembly and testing process. The study compares the results between the conventional and outlier analysis. Findings: The conventional style of checking and approving the engines based on the value pattern of Specific Fuel Consumption (SFC) which uses the design specification comparison with the actual data generally yields very minimal scope for the improvement of product quality in the perspective of design, Safety and reliability of the product because of the adherence of the same design specifications of the part drawings supplied by various suppliers. The competitive automotive manufacturing domain demands a different approach with which a better scope could be identified towards the improvement of product quality which is undoubtedly data mining. The outlier analysis using inter quartile range on the sample data of 500 engines revealed many important aspects where the improvement scope for quality has been identified as 15,000 Parts Per Million (PPM) against400 PPM of conventional quality analysis for the same data. Improvement/Application: This research is to offer an inclusive model hypothetically with actual data both for engineering and management people of manufacturing domain about the insights and benefits of employing data mining techniques towards the improvement of product quality with proven results.