Posts in category Terms

A mathematical method that employs probability theory for inferring the properties of a population parameter from which the sample is taken is known as inferential statistics. Inferential statistics is a set of methods used to make generalizations, estimations, or predictions. Let’s say that we want to determine the statistical capability of a process. And, let’s say that 28 machines are producing a particular part. This process has 14 operators that run the parts. The operators use their own micrometers to measure key characteristics of the parts. There are three sources of raw materials that feed into the process. The process runs two shifts per day five days a week.

A hypothesis test is a method for making rational decisions about the reality of effects. Most decisions require choosing from one or more alternatives. The decision is based on incomplete information. A team might be considering using a different method which they believe will give them a better result. Their theory is that method A is going to be better than method B.

Pioneered by the lean guru Taiichi Ohno, referred to load leveling as “heijunka” (HIGH JUNK AH), meaning a technique for reducing the mode of (waste). Load leveling is a method for reducing large fluctuations in customer demand.

Lean guru Taiichi Ohno can be quoted, “The slower, but consistent tortoise causes less waste and is much more desirable than the speedy hare that races ahead and then stops occasionally to doze. The Toyota Production System can be realized only when all the workers become tortoises.” You could have leveling by volume, or leveling by product. Leveling by volume: A family of products goes through essentially the same process steps.

Dr. Genichi Taguchi, a Japanese quality consultant and engineer, came to the United States in the 1980s concentrates on loss. In conjunction with this view of quality, he developed the idea of a loss function. If we think of quality in a competitive way, we should view that there is a target value (i.e., nominal-is-best, larger-is-best, or smaller-is-best). When measuring an end-product-parameter, any departure away from the designer’s intended target creates a loss to society, or a loss to the producer. Either way, ultimately the customer ends up paying for it.

The kano model classifies product or service attributes based on the perception of the voice of the customer. There are three classifications (basic needs, performance, and excitement) which are useful for guiding design decisions in that they are an indication of when good is good enough, and when more-is-better.

When we are trying to determine the relationship between two variables, one of the relationships might be the equation of a straight line [i.e., y = (f)x.] With the least squares method, the team is using the linear equation. The linear equation represents the points found on the scatter diagram. In essence, the team is using the least squares criterion, meaning that the line fitted to the paired data points must be such that the sum of the squares of the vertical distances from the points to the line is as small as possible.

Levene’s test is used to access variation homogeneity. Levern’s test is useful for parametric tests such as the T-test and Analysis of Variance (ANOVA) that require a precondition of homogeneity.

A matrix diagram is a tool that illustrates the critical relationships between or among two or more groups. It can be enhanced to show things like the strength of the relationship or the different aspects of the relationship.

A defect location check sheet (also known as a defect map or a measles chart) is a structured, prepared form for collecting and analyzing data that provides a visual image of the item being evaluated so that data can be collected visually rather than merely collecting a count of the number of defects.

Measurement error is the difference between the actual value of a characteristic being measured and the value obtained by the measurement system. Given that the measurement device is discriminate (sensitive to variation of the product being measured), with each repeat of a measurement, there will be variation. By the way, if no variation is seen, measure with a more discriminate device. There is systemic error and random error. The systemic error occurs when we use the measurement device in the same way and in the same condition. The random error will vary from reading to reading. There can also be variation due to the human element (when applicable).