Overview for Individual Distribution Identification

Use Individual Distribution Identification to identify an appropriate distribution or transformation for your data before you perform an analysis.

Many statistical analyses, such as capability analysis, are based on the assumption that your data follow a particular distribution. Individual Distribution Identification provides probability plots and goodness-of-fit tests that allow you to do the following:

• Determine which of 14 distributions provides the best fit for your data.
• Determine whether a transformation is effective to fit your data to a normal distribution.

For example, a floor tile manufacturer wants to assess the capability of a manufacturing process to produce tiles that are not warped. The quality analyst measures warping in random samples of tile and records the data. Because the distribution of the warping data is not known, the analyst uses individual distribution identification and determines that the Weibull distribution provides the best fit. Therefore, the analyst decides to evaluate the capability of the process using a nonnormal capability analysis based on the Weibull distribution.

You can also use individual distribution identification to transform your data using a Box-Cox or Johnson transformation and to store the transformed data values in the worksheet for further analysis. For example, a quality analyst needs to perform several statistical analyses that are based on the assumption of normality. When a normality test reveals that the sample process data are not from a normal distribution, the analyst uses individual distribution identification to transform the data using a Box-Cox and a Johnson transformation, to evaluate the effectiveness of each transformation, and to store the transformed values for further analysis.