Process Metrics

Process metrics need to lead to the most appropriate behaviors. Processes have variability and may or may not have specifications.

Performance measurements for processes need to provide direction to the most appropriate behaviors considering both process variability and any specification that may exist. The output of processes can have both common-cause variability and special-cause variability.

In process metric reporting, typical process variability is separated from unusual events or trends. Traditionally this separation is make using statistical process control (SPC) charts such as x-bar and R charts and p-charts. How a process is performing for an in-control process relative to specifications traditionally involves techniques such as process capability indices.

However, traditional control charting and process capability reporting have mathematical issues. An Integrated Enterprise Excellence (IEE) 30,000-foot-level reporting format addresses these issues. IEE 30,000-foot-level reporting provides both a process stability assessment and predictive statement for stable processes in one chart.

Statistical Transformations for Normality: NOT Transforming The Data Can Be Fatal To Your Analysis

Performance Measurements, Process Metrics /

There is debate whether, in statistical process control (SPC), a data transformation should be considered when constructing an individuals chart. This article shows, using real data, why an appropriate data transformation is very important to determine the best action or non-action to take in both manufacturing and transactional processes at any point in time. Described in this article is also an enhancement to traditional process control charting methodology. The described statistical business performance charting (SBPC) system can, for example, reduce firefighting when the approach replaces organizational goal-setting red-yellow-green scorecards, which often have no structured plan for making improvements. In addition, the methodology provides predictive performance statements. Donald Wheeler and Forrest have a difference of opinion about the need to transform data when a transformation makes physical sense. The reason for writing this article is to provide information on the reasoning for Forrest’s position. Hopefully this supplemental explanation will provide readers with enough insight so that they can make the best logical decision relative to considering data transformations or not.

Statistical Transformations for Normality: NOT Transforming The Data Can Be Fatal To Your Analysis Read More »

Log Transformation: Non-Normal Data, To Transform Or Not To Transform

Performance Measurements, Process Metrics /

This article was written as an alternative approach to analyzing non-normal data to that which was presented by Dr. Don Wheeler in a previous Quality Digest newsletter. This article illustrates, from a high level, or 30,000-foot-level, when and how to apply transformations and present results to others so that the data analysis leads to the most appropriate action or nonaction. Statistical software makes the application of transformations simple.

Log Transformation: Non-Normal Data, To Transform Or Not To Transform Read More »

Enhancing Process Capability Indices Cp and Cpk Reporting

Performance Measurements, Process Metrics /

Measurement issues can be prevalent at all levels of an organization. To add to this dilemma, the basic calculation and presentation of metrics can sometimes be deceiving. Organizations often state that suppliers must meet process capability objectives, typically measured in Cp, Cpk, Pp and Ppk. The requesters of these objectives often do not realize, however, that these reported numbers can be highly dependent upon how data is collected and interpreted. Also, these process capability metrics typically are utilized only at a component part level. To resolve these issues, practitioners need a common, easy-to-use fundamental measurement for making process stability and capability assessments at all levels of a business, independent of who is making the assessment – something beyond Cp, Cpk, Pp and Ppk.

Enhancing Process Capability Indices Cp and Cpk Reporting Read More »

Understanding and Enhancing Process Capability Cpk Reporting

Performance Measurements, Process Metrics /

Cpk is one of many capability metrics that are available. When capability metrics are used, organizations typically provide a Cp and a Cpk. In this paper we will discuss the mechanics of these two capability metrics, along with the pros and cons. In summary, the Cpk can provide insight on performance to a requirement if the process data used in the calculation comes from a normal distribution. If the process data is non-normal or it is the result of a combination of processes (a mixture of processes) then it provides an underestimation of the true non-conformance capability.

Understanding and Enhancing Process Capability Cpk Reporting Read More »

Process Capability Analysis Cpk: Understanding and a Better Way

Performance Measurements, Process Metrics /

Have you always liked capability metrics, such as Cp and Cpk, and wondered what they really said? Well maybe you are not loosing sleep over this issue, but like many statistics, capability metrics are used without a full understanding of their message. View this webinar and continue your process improvement knowledge as we discuss each of these three questions. We will focus on the application rather than the theory of these capability metrics. Yes, there will be equations and such, but only in regards to explaining the concepts.

Process Capability Analysis Cpk: Understanding and a Better Way Read More »

Control Chart for Non-Normal Data Example: Log-normal Distribution with Negative Values

Performance Measurements, Process Metrics /

This paper addresses 3-parameter log-normal distribution control charting. Statistical Process Control (SPC) has a primary purpose, which is to identify when special cause conditions occur for timely corrective actions. SPC textbooks and training state that an individuals control chart (X) or an individuals and moving range chart (XmR) control chart is appropriate and should be used when tracking individuals data. With an individuals chart, between subgroup variability affects the magnitude of the control limits from the overall response mean. This paper illustrates how the individuals control chart is not robust to non-normality. When data are not normally distributed (e.g., when there is a boundary condition) the data need a normalizing transformation that is appropriate for the process being evaluated. The log-normal distribution often is a good fit for these situations. However, the log-normal distribution cannot accept negative numbers, which are often needed to describe this type of process response. This paper describes the use of the 3-parameter log-normal distribution in 30,000-foot-level control charting. In this 30,000-foot-level metric reporting methodology, process response is evaluated for regions of stability. Within identified stable regions, a process capability non-conformance estimate can then be reported if a specification exists. If there is a recent region of stability, one can consider the data in this region to be a random sample of the future; hence, a prediction statement can be made. An enterprise can assess its value-chain metrics collectively – where each has 30,000-foot-level reporting – to determine where improvements can be made that positively impact the enterprise financials as a whole. Goals to these metrics would pull for a process improvement or design project creation that positively impacts these 30,000-foot-level metrics.

Control Chart for Non-Normal Data Example: Log-normal Distribution with Negative Values Read More »

Enhanced Control Chart for Count Data with Prediction Process Capability Statement

Performance Measurements, Process Metrics /

This paper addresses infrequent failure data control charting. A primary purpose of Statistical Process Control (SPC) is to identify when special cause conditions occur for timely corrective actions. SPC textbooks and training state that a c-chart or u-chart should be used for count data tracking. With a c-chart, the tracked response is the number of counts, which has time-series identified rational sub-groups. A u-chart has a similar tracking but only occurrence rate is track. Described in this paper are technical issues that make the c-chart and u-chart ineffective, especially when the counts (e.g., safety incidents) are very low for the subgrouping (e.g., months). The technical reason for this occurrence is discussed along with an alternative 30,000-foot-level reporting system that addresses these issues. Traditional c-chart and u-chart control charts can lead to inappropriate activities, since the underlying assumptions for these charts are often not valid in the real world. In the 30,000-foot-level metric reporting methodology, which centers on use of the individuals control chart, process response is evaluated for regions of stability, where time between incidents is a response that can be tracked for adding power to the test when there are infrequent count occurrences. Within identified stable regions, a process capability estimate can then be reported. If there is a recent region of stability, one can consider the data in this region to be a random sample of the future; hence, a prediction statement can be made. An enterprise can assess its value-chain metrics collectively – where each has 30,000-foot-level reporting – to determine where improvements can be made that positively impact the enterprise financials as a whole. Goals to these metrics would pull for a process improvement or design project creation that positively impacts these 30,000-foot-level metrics.

Enhanced Control Chart for Count Data with Prediction Process Capability Statement Read More »

Enhanced Control Charts for Defective Proportions with Predictive Process Capability Statement

Performance Measurements, Process Metrics /

This paper addresses technical problems with p-chart and provides an alternative. Statistical Process Control (SPC) has a primary purpose of identifying when special cause conditions occur for timely corrective actions. SPC textbooks and training state that a p-chart should be used for non-conformance rate tracking, when the data are attribute. With a p-chart, there is non-conformance rate reporting that has time-series identified rational sub-groups. Described in this paper are technical issues with the p-chart that can lead to a falsely identifying common cause process variation as though it were special cause. The technical reason for this occurrence is discussed along with an alternative 30,000-foot-level reporting system that addresses these issues. Traditional p-chart control charts can lead to much firefighting, since the underlying assumptions for these charts are often not valid in the real world. In the 30,000-foot-level metric reporting methodology, which centers on use of the individuals control chart, process response is evaluated for regions of stability. Within identified stable regions, a process capability non-conformance estimate can then be reported. If there is a recent region of stability, one can consider the data in this region to be a random sample of the future; hence, a prediction statement can be made. An enterprise can assess its value-chain metrics collectively – where each has 30,000-foot-level reporting – to determine where improvements can be made that positively impact the enterprise financials as a whole. Goals to these metrics would pull for a process improvement or design project creation that positively impacts these 30,000-foot-level metrics.

Enhanced Control Charts for Defective Proportions with Predictive Process Capability Statement Read More »