Upload
libyaflower
View
220
Download
0
Tags:
Embed Size (px)
Citation preview
Statistics and DOE
David M. Lyth, Ph.D., C.Q.E.Department of Industrial & Manufacturing EngineeringWestern Michigan University
Managing by facts
“In God we Trust—All others bring data”
ISO9000 AuditorISO9000 Auditor
Copyright, Lyth, WMU
Data and Quality
� What are the most pressing problems in our department?
� What do defects cost?What causes most defects?� What causes most defects?
� What is the defect removal rate or quality engineering?
� How many defects are severe?� What is the average cost of repairing a
defect?
Copyright, Lyth, WMU
Statistical Thinking
�All work occurs in a system of interconnected processes
�Variation exists in all processes�Variation exists in all processes�Understanding and reducing
variation are the keys to success
Copyright, Lyth, WMU
Sources of Variation in Production Processes
Materials Operators Methods MeasurementInstruments
Tools HumanInspectionPerformance
EnvironmentMachines
INPUTS PROCESS OUTPUTS
Copyright, Lyth, WMU
Variation
�Many sources of uncontrollable variation exist (common causes)
�Special (assignable) causes of �Special (assignable) causes of variation can be recognized and controlled
�Failure to understand these differences can increase variation in a system
Copyright, Lyth, WMU
Problems Created by Variation�Variation increases unpredictability. �Variation reduces capacity utilization. �Variation contributes to a “bullwhip”
effect. effect. �Variation makes it difficult to find root
causes. �Variation makes it difficult to detect
potential problems early.
Copyright, Lyth, WMU
Two Fundamental Management Mistakes
1. Treating as a special cause any fault, complaint, mistake, breakdown, accident or shortage when it actually is due to common causesdue to common causes
2. Attributing to common causes any fault, complaint, mistake, breakdown, accident or shortage when it actually is due to a special cause
Copyright, Lyth, WMU
Using data
�Managing a process�Understanding a process�Controlling a process� Improving a process
Copyright, Lyth, WMU
Collecting data-questions to answer�What is the purpose of collecting data?�What data do I need?�How will I collect it?�How will I analyze it?�How will I analyze it?�What conclusion am I attempting to
draw?�How will I present those conclusions to
achieve my objectives?
Copyright, Lyth, WMU
Collection problems� Data accuracy� Representative data� Time series data� False data� False data� Mistaken data� Incorrect data� Unavailable data� Useless data
Copyright, Lyth, WMU
Ways to classify data
�Time�Time to failure�Development phase �Team creating defect�Activity producing error�Function of the defect�Severity of the defect
Copyright, Lyth, WMU
Possible causes
�Communication – “I didn’t understand what I was supposed to do.”
�Education – “I didn’t know how to do it”Transcription – “I was doing it right, but �Transcription – “I was doing it right, but something went wrong in the process.”
�Oversight – “I just forgot.”
Copyright, Lyth, WMU
Statistics
�Descriptive statistics� Inferential statistics�Estimation�Hypothesis testing�ANOVA�Regression
Copyright, Lyth, WMU
Data fitting
Goodness of fit test
Χ2 = Σ (fo-fe)2/fe
Ho: Data fits particular distribution
(specify parameters)
H1: Data does not fits particular
distribution (specify parameters)Copyright, Lyth, WMU
Statistical Methods
�Descriptive statistics�Statistical inference�Predictive statistics�Predictive statistics
Copyright, Lyth, WMU
Statistical Foundations
�Random variables�Probability distributions�Populations and samples�Populations and samples�Point estimates�Sampling distributions �Standard error of the mean
Copyright, Lyth, WMU
Descriptive statistics
�Measures of central tendency (mean, median, mode)
�Measures of dispersion (variance, standard deviation, range)standard deviation, range)� Histogram� Pie chart� Frequency curve
Copyright, Lyth, WMU
Inferential statistics
�Estimation�Hypothesis testing�ANOVA�Regression
Copyright, Lyth, WMU
Important Probability Distributions�Discrete
�Binomial�PoissonPoisson
�Continuous�Normal�Exponential
Copyright, Lyth, WMU
Central Limit Theorem
� If simple random samples of size n are taken from any population, the probability distribution of sample means will be approximately normal as nbecomes large.
Copyright, Lyth, WMU
Statistical Tools
Copyright, Lyth, WMU
Sampling Methods
�Simple random sampling�Stratified sampling�Systematic sampling�Systematic sampling�Cluster sampling�Judgment sampling
Copyright, Lyth, WMU
Sampling Error
�Sampling error (statistical error)�Nonsampling error (systematic
error)error)�Factors to consider:
�Sample size�Appropriate sample design
Copyright, Lyth, WMU
Excel Tools for Statistics
� Tools…Data Analysis… Descriptive Statistics
� Tools…Data Analysis…Histogram
Copyright, Lyth, WMU
Enumerative and Analytic Studies�Enumerative study – analysis of a static
population�Analytic study – analysis of a dynamic
time seriestime series
Copyright, Lyth, WMU