Page 128 - Solid Waste Analysis and Minimization a Systems Approach

106 THE SIX SIGMA SYSTEMS APPROACH FOR DEPLOYMENT

6.3 Measure

The measure phase is to identify correct measures, establish a baseline, and eliminate
trivial variables (Gupta et al., 2007). The following tools are important to understand:

■ Basic statistics
■ Statistical thinking
■ Cost of quality
■ Measurement system analysis
■ Critical parameters
■ Critical to quality

There are two types of statistics, descriptive and inferential. The descriptive statis-
tics summarize the historical data (Gupta et al., 2007). Basic descriptive statistical
analysis consists of the mean, median, mode, range, variance, and standard deviation
(Gupta et al., 2007). Also measuring the cost of quality is critical because high varia-
tions and inconsistencies can cause high cost and waste valuable resources. Inferential
statistics is based on analysis of the sample to infer performance of the process (Gupta
et al., 2007). It is generally concerned with the source of the data and seeks to make
generalizations beyond the data at hand. Inferential statistics can include regression
analysis and hypothesis testing among many others.

6.4 Analyze

The analysis phase begins the convergence of possibilities toward the root cause of the
problem (Gupta et al., 2007). Key analysis tools consist of multi-vary analysis, cause
and effect diagrams, regression analysis, and failure modes and effects analysis (FMEA).
The cause and effect diagram is used to identify the source of the problem of the
process. The cause and effect diagram shows each branch of the process and the inputs
related. From the diagram one is then able to notice the different contributors to the
problem or process.
The regression analysis helps to identify causes and show correlations between vari-
ables. Often when solving problems involving many variables most have an inherent rela-
tionship. Regression is a way to estimate variable y if there is x amount of variable z.
There are independent and dependent variables. Dependent variables are the responses
and the independent variables are the causes. Relationships are not always deterministic;
that is, x does not always give the same value for y (Walpole and Myers, 1993). As a result
relationships cannot be exact, and regression methods are used to predict relationships.
Multi-vary analysis is a way to reduce the number of potential causes by removing
the trivial ones (Gupta et al., 2007). Many data sets have hidden or not easily recognized
similarities, patterns, or structures (Graham, 1993). Different techniques that can be
used to address this are cluster analysis, two-factor analysis of variance, and three-factor
experiments.