# Patterns in Data using Minitab

Document Sample

```					Six Sigma
Patterns in Data Minitab

Patterns in Data
• Six Sigma heavily reliant on Data:
– Several techniques, systems used to store, retrieve, analysis of data
• Data Warehousing: Comprised of Several Components
– – – – – – Source Systems: Where Data comes from Transport and Cleansing: Moving Data Central Repository: Main location for Data Metadata: Description of what data is available Data Marts: Easy access for end users Operational Feedback: Integration of decision support to operations

Patterns in Data
• OLAP (On-Line Analytic Processing): Allows end users to extract information from large databases
– Uses concept of cubes where data can be arranged to give users information required.

• Data Mining: Used to provide Patterns in Data via automatic or semi-automatic means

– It is helpful for Six Sigma practitioners to have access to these types of tools. – In cases where data collection is manually collected: Identifying Patterns consists of Plotting the Data, and then looking at Descriptive Statistics.

Patterns in Data
• The three rules for Data Analysis:

Patterns in Data
• The three rules for Data Analysis:
– Plot the Data

Patterns in Data
• The three rules for Data Analysis:
– Plot the Data – Plot the Data

Patterns in Data
• The three rules for Data Analysis:
– Plot the Data – Plot the Data – Plot the Data

• Do a Histogram • Do a Box Plot • Check for Normality

Histogram

Histogram

Stem and Leaf Plot

Box Plot

Test for Normality

Test for Normality

Test for Normality
• Note P value: If the p value is low, the Null Hypothesis must go. • Null Hypothesis in this case is rejecting the assumption of normality. • If P value is less than .05 then the Null Hypothesis is rejected • Simple Way to Remember: If P value is greater than .05, then assume data is normal.

Patterns in Data
• What Happens if Data is not Normal? (Pyzdek)
– Do nothing: Often the histogram or normality plot shows that normality exists “where it counts”. – Transform the Data: Box-Cox

Patterns in Data
– Use Averages: Averages of subgroups always tend to be normally distributed. – Fit to another distribution:
• Use Goodness of Fit tests to find proper Distribution

– Use a non-parametric technique which do not make any assumptions about underlying distribution of data.

Stratification
• Used when data comes from several different sources and have been “lumped” together. • Example: Data Collected from several different machines making the same part. • The scatter diagram (units produced over time) show no pattern. • Using Stratification showing data by machine can help in identification of patterns.

Stratification
Output
16 14

12

10

8

6

4

2

0 12/30/2008 1/1/2009 1/3/2009 1/5/2009 1/7/2009 1/9/2009 1/11/2009 1/13/2009 1/15/2009 1/17/2009 1/19/2009 1/21/2009

Stratification
7
6

5

4 Machine A Machine B 3 Machine C

2

1

0 12/30/2008 1/1/2009 1/3/2009 1/5/2009 1/7/2009 1/9/2009 1/11/2009 1/13/2009 1/15/2009 1/17/2009 1/19/2009 1/21/2009

Measure Deliverables
• Process Map
– Identification of Inputs, Process, and Outputs

• Documented Data Collection Plan
– Measurement System – Operational Definitions

• • • •

Run Charts, Control Charts of Process Distribution of Data Possible Yields through-out Process Patterns in Data

```
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 237 posted: 10/1/2009 language: English pages: 20
Description: Using Minitab to look at Box Plots, Stratification to aid in finding patterns in data.