Assessing Supplier’s Process Capability using Truncated Normal Distribution Data

Process Capability Indices (PCIs) have been proposed to assess process capability in real world application over the past four decades. Today, purchasing personnel also uses PCIs to select best supplier. To select best supplier using supplier’s values of PCIs may not be reliable due to fear of data manipulation. To assess supplier’s process capability, purchasing personnel needs process distribution with parameters referring received samples from supplier. Most of the time, received samples conform hundred percent to the specification, because products are categorized as conforming and nonconforming by supplier before being sent to customer and only conforming products are sent to customer. If the process distribution is normal then process distribution identified by referring to received sample products is truncated normal. From a customer’s point of view intention of this paper is estimation of process distribution parameter referring to truncated normal data by identifying best method of parameter estimation with respect to accuracy and precision of estimates. It is found that method of moments provides best estimators of process parameters without loss of efficiency as compared to other competing methods. Through simulation, performance of method of moments is compared with other competing recently developed methods which include maximum likelihood estimation starting from re-parameterization and quantile-filling algorithm (QA)based on EM (Expectation-Maximization) algorithm. Estimated parameters are used to estimate supplier’s process capability using probability based PCIs through illustrative example.