1 Minitab Basics An Introduction Beauchamp Consulting Proprietary
1 Minitab Basics - An Introduction Beauchamp Consulting – Proprietary and Confident
Minitab Session - Starting Point 2 Menu Bar Session Window: • Analytical Output Data Window: • A Worksheet, not a Spreadsheet Beauchamp Consulting – Proprietary and Confident
Minitab Session - Entering Data 3 Menu Bar Session Window: • Analytical Output Data Window: • Each row shows one particular data point • Each column describes a characteristic of that particular data point. Beauchamp Consulting – Proprietary and Confident
Minitab Session - Graphs 4 Command in Minitab: Graphs>Pie Chart> enter “Pet” into cell titled “Chart Data In: ”>OK Note that Minitab will automatically put the columns in the left box, all you have to do is double click the right column, you don’t have to type in “Pet”. Beauchamp Consulting – Proprietary and Confident
Minitab Session - Graph Window 5 Session Window Data Window Project Window Graph Window Beauchamp Consulting – Proprietary and Confident
Minitab Windows - Summary Data Window q q q Data window is a worksheet, not a spreadsheet. Variables are usually entered in columns, observations in rows. Each column has a title area - 31 characters or less. Everything in a column is considered to be the same variable. Can be accessed using Data Window button, Ctrl-Tab, Ctrl-D, or via the Windows menu. Graph Window q Generated whenever you run a command that results in a graphical output. Session Window Stores numeric output and character graphs that result from executing a statistics command. q Can be accessed using Session Window button, Ctrl-Tab, Ctrl-M, or via the Windows menu. Info Window q q q 6 q History Window Stores synopsis of your data window. Stores information about columns, constants and matrices. Can be accessed using Ctrl-Tab, Ctrl-I, or via the Windows menu. Stores all commands initiated from either the menus or the session window. q Can be accessed using Ctrl-Tab, Ctrl-H, or via the Windows menu. q Beauchamp Consulting – Proprietary and Confident
Minitab Menus - Summary 7 The File menu is very similar to other Windows packages. From the File menu, you can print and save the window that is currently active. The File menu options change depending on the window that is currently active. q When you save a project, you save all the worksheet, session, info, history and graph windows that are currently. q File > Open Worksheet allows you to open formats other than. mtw (MINITAB) files, such as Excel (. xls) files. q q File Menu Edit Menu q The Edit menu is similar to the Edit menu in most standard Windows applications. Data Menu & Calc Menu q The Manip and Calcu menus are used to move data around or to create new variables automatically. q The Stat menu contains the statistical commands that you will use during your project. q The Graph menu contains the commands that you will use to do graphical analysis during your project. Stat Menu Graph Menu Beauchamp Consulting – Proprietary and Confident
Minitab Help Menu 8 Minitab has a comprehensive help system with detailed documentation of all features, complete with examples of how all the menu commands are used, and how to interpret graphical and statistical output which result from the use of the commands. Beauchamp Consulting – Proprietary and Confident
Minitab Data Window (Worksheet) - Tips 9 MINITAB’s worksheet has characteristics which differ from a spreadsheet. Within the worksheet: q ANYTHING in a single column is considered part of the same variable. q Column names are ABOVE the column, not in the first row. q If the value you enter into the first row is a number, Minitab assumes that the entire column will contain numbers (N). q If the value you enter into the first row is something other than a number or date, Minitab assumes that the entire column will be text (T). Minitab tells you this by placing a ”T” beside the column number. Date/Time data will have a “D” next to the column number. q. If you have mistakenly converted a column to text, simply convert the column back to numeric through: Manip > Change Data Type > Text to Numberic q. Fill in dialog boxes to indicate column to be changed and where to store it (may be stored in same location). q. Manip q > Change Data Type can be used for changing between all three data types. The orientation of the data entry arrow determines whether the ENTER key will take you to the next row or the next column. Beauchamp Consulting – Proprietary and Confident
Minitab Data Window (Worksheet) - Tips (cont. ) 10 MINITAB’s worksheet has characteristics which differ from a spreadsheet. Within the worksheet: q Rows (or Columns) may be inserted with the following command: Editor > Insert Rows (Columns) q. Rows (Columns) will be inserted above the highlighted cell. If you want to insert x number of rows (columns), highlight x number of rows (columns) prior to the insert new row (column) command q To delete rows (columns), highlight them and select: Edit > Delete Cells. The other rows (columns) will be moved up (or over). q To erase the contents of rows (columns), highlight them and select: Edit > Clear Cells. The other rows (columns) will NOT be moved. q Minitab cells contain values only, not formulas, like Excel. To perform mathematical funcitons, use: Calc > Calculator. q Control E returns you to the last dialog box. Beauchamp Consulting – Proprietary and Confident
Importing Data into Mintab - Tips q Structure the data so that each variable is in a single column. q Each column must have a title. q The column title must have fewer than 31 characters and be in a single cell. q All data must immediately follow the column title (no empty row). q Do not put empty rows between rows of data. q Columns containing dollar signs or commas cannot be transferred to Mintab using Copy or Paste. To use Copy or Paste, you must first reformat these numbers to include only decimal points. Columns with dollar signs and commas will successfully import as generic number. 11 Beauchamp Consulting – Proprietary and Confident
Minitab Worksheet with Errors 12 Errors q q q Header of “Sales 1999” needs to be removed Should not have empty rows Should not have dollar signs Should not have commas Should not have a total at the bottom Beauchamp Consulting – Proprietary and Confident
Minitab Worksheet without Errors 13 Beauchamp Consulting – Proprietary and Confident
Using Minitab Output 14 Copying Graphs 1. Make sure the graph window is active in Minitab 2. Click the right mouse button and select Copy Graph (or choose Copy from the Edit menu) 3. Open the file that you want to paste the graph into, e. g. a Power Point file. 4. Paste the graph to the file using the Ctrl-V command or by choosing Edit > Paste Saving your Work From the Minitab File menu you have several save options. 1. Save Project - this will save all of your output in the various windows (graphs, worksheet, etc. ) 2. Save Current Worksheet - this will only save the active worksheet Finding your Output From the Window menu 1. Scroll to the bottom of the window menu 2. You will see all graphs, sessions, worksheets you have open. Select the one you want to show. 3. If you select the tile option – Minitab will show all windows open in a ‘tiled’ format on your screen. Printing your Output 1. Make sure the window that you want to print is active in Mintab 2. From the File menu, choose Print Beauchamp Consulting – Proprietary and Confident
15 Minitab “Cheat Sheets” Beauchamp Consulting – Proprietary and Confident
Measure Cheat Sheet What do I want to know? Tool 16 Command in Minitab How is my data distributed? What is the mean, std deviation, median, etc. Basic Stats Stat>Basic Stats>Graphical Summary or Display Descriptive Stats (data without graphs) Is my data normally distributed? Normality Plot Stat>Basic Stats>Normality Test How is my data distributed over time? Run Chart Stat>Quality Tools>Run Chart Are there any relationships between Y and my Xs? Discrete X and Y Pareto Are there any obvious relationships between Y and my Xs? Discrete X, Continuous Y Box. Plot Graph>Box Plot Are there any relationships between Y and my Xs? Continuous X, Continuous Y What is the strength of the relationship between X&Y? Scatter Plot Graph>Scatter Plot Pearson Product Moment correlation coefficient Stat>Basic Stats>Correlation Are there any relationships between Y and my Xs? Discrete Xs, Continuous Y (Multiple Xs) Multivari Plot Stat>Quality Tools>Multi-Vari Chart Check for p>. 05 to prove it’s normal Check for evidence of special causes Stat>Quality Tools>Pareto Chart If you have 1 column of data, use “Chart Defects data in”, if you have a table of data summarizing defect counts by defect type, use “Chart Defects Table”. (use two columns of data, X in one column, Y in the other) Beauchamp Consulting – Proprietary and Confident
Measure Cheat Sheet What do I want to know? Tool Y = f(x) Y = ax 1 + bx 1 + cx 1 + dx 1 + … 17 Command in Minitab and terminology Y = Dependent Variable X = Independent Variable How capable is my process of producing defect-free products/services? Continuous Data Process Capability Six Sigma> Process Report How capable is my process of producing defect-free products/services? Discrete Data Process Capability Six Sigma> Product Report Best in Class – Best this process can do Sigma Value Y = ax 1 + bx 1 + cx 1 + dx 1 + … Short Term Capability sshort-term = Zshort-term Actual performance over long term – in Z terms Y = ax 1 + bx 1 + cx 1 + dx 1 + … Long Term Capability slong-term = Zlong-term Level of Control of Process – how much does Z shift and drift between short and long term? Y = ax 1 + bx 1 + cx 1 + dx 1 + … Shift sshift= Zshift sbenchmark = Zbenchmark Beauchamp Consulting – Proprietary and Confident
Analyze Cheat Sheet What do I want to know? Tool 18 Command in Minitab or Excel What are potential Xs driving performance in this Y? Cause & Effect Diagram Stat>Quality Tools>Cause and Effect Diagram What affect do the various steps have on overall cycle time and what percentage are truly value adding? Cycle Time/Value Analysis Spreadsheet Value added cycle time analysis spreadsheet. xls What are some potential Xs that represent a high likelihood to fail in this process/service? FMEA: failure Mode & effects Analysis Fmeaform(1). xls Is this a significant X – does it cause a shift in Y’s mean? Y: continuous, X: discrete (one X) Hypothesis Testing: 2 sample t test Stat>Basic Stat>2 sample t Is this a significant X – does it cause a shift in Y’s mean (m)? Y continuous, X: discrete (more than one X) Hypothesis Testing: One way ANOVA Stat>ANOVA>One-Way Is this a significant X – does it cause a shift in Y’s variation (s)? Y continuous, X: discrete Hypothesis Testing: Homogeneity of Variance Stat>Basic Stat>2 variances Check for p<0. 05 to prove ma not equal to mb not equal to mc … at least one m is different. Check for p<0. 05 to prove at least one variance is different Beauchamp Consulting – Proprietary and Confident
Analyze Cheat Sheet What do I want to know? How does this performance of Y compare with a target (Baseline, performance Standard, etc. ) Tool 1 -Sample t test Is this a significant X – does it Chi-Square cause a difference in proportion in Test of Y? Y: discrete, X: discrete Independence Minitab Heading: Chi Square Set-up in Minitab Population 1 19 Command in Minitab Stat>Basic Stats>1 sample t Check for p<0. 05 to prove ma not equal to the target value Stat>Tables>Chi Square Test Check for p<0. 05 to prove the proportions of the populations are different – recommend use setup shown below for table input in Minitab Population 2 Defect Count of defects in Population 1 Count of defects in Population 2 Non Defects Count of good units in Population 1 Count of good units in Population 2 Beauchamp Consulting – Proprietary and Confident
Analyze Cheat Sheet What do I want to know? Tool 20 Command in Minitab Is this a significant X – does it Hypothesis cause a shift in Y’s mean? Y: Testing: continuous, X: continuous or discrete ANOVA GLM (More than one X) Stat>ANOVA>General Linear Model Is this a significant X – does it correlate linearly with Y? (every time X moves a certain amount, Y moves in a linear fashion along with X) Y and X continuous: 1 X Simple Linear Regression Stat>Regression Is this a significant X – does it correlate linearly with Y? (every time X moves a certain amount, Y moves in a linear fashion along with X) Y and X continuous: (More than one X) Multiple Regression Stat>Regression Is this a significant X – does it correlate linearly with Y? (every time X moves a certain amount, Y moves in a linear fashion along with X) Y discrete and X continuous Binary Logistic Regression Stat>Regression>Binary Logistic Regression Check for p<0. 05 to prove at least one m is different Check for p<0. 05 to prove there is a linear relationship between Y and X … the model is a good predictive fit Check for p<0. 05 to prove there is a linear relationship between Y and X’s … the model is a good predictive fit Check for p<0. 05 to prove there is a linear relationship between Y and X … the model is a good predictive fit Beauchamp Consulting – Proprietary and Confident
A 10. 1 - Hypothesis Testing 21 6. Choose the appropriate tool to test the validity of your statement Minitab Tool Stat>Basic Stat>2 Sample t Stat>ANOVA>One-Way… Stat>ANOVA>General Linear Model Stat>Regression>Regression Stat>Tables>Chi Square Test Stat>Regression>Binary Logistic Regression Stat>Basic Stats>2 variances Stat>Basic Stats>1 sample t Beauchamp Consulting – Proprietary and Confident
Improve Cheat Sheet What do I want to know? Tool 22 Command in Minitab or Excel Is there a configuration of Xs that will produce the best performance in Y? DOE: set-up design State>DOE>Factorial>Create Factorial Design What does the DOE prove about the Xs as they relate to Y? DOE: analysis Stat>ANOVA>Main Effects Plot Do any of the Xs interact to affect performance of Y? DOE: analysis Stat>ANOVA>Interactions Plot How good is the DOE model? How well can we fit the data to the equation? DOE: analysis Stat>DOE>Factorial>Analyze Factorial Design What are the risks inherent in implementing this solution? Risk Analysis Riskassessmenttool. xls Check for p<0. 05 to prove there is a significant main effect Beauchamp Consulting – Proprietary and Confident
Control Cheat Sheet What do I want to know? Upper Control Limit: 3 s on either side of the mean Tool Control Charts 23 Command in Minitab and terminology Established by project team by taking 20 subgroups of data, 3 -5 data points in each subgroup Defect refers to the failure to meet one of the acceptance criteria. Defective refers to an entire unit that fails to meet acceptance criteria regardless of the number of defects on the unit. Is my data Defect or Defective? Is my process in control? Y is discrete, measuring defects, with constant sample size >5 in subgroup Control Charts c Chart Stat>Control Charts>Attributes Charts>c Is my process in control? Y is discrete, measuring defects, with a variable subgroup sample size Control Charts u Chart Stat>Control Charts>Attributes Charts>u Is my process in control? Y is discrete, measuring defectives, with a constant subgroup sample size usually >49 Control Charts np Chart Stat>Control Charts>Attributes Charts>np Is my process in control? Y is discrete, measuring defectives, with a variable subgroup sample size usually >49 Control Charts p Chart Stat>Control Charts>Attributes Charts>p Beauchamp Consulting – Proprietary and Confident
Control Cheat Sheet What do I want to know? Tool 24 Command in Minitab and terminology Upper Control Limit: 3 s on either side of the mean Control Charts Established by project team by taking 20 subgroups of data, 3 -5 data points in each subgroup Is my Y in control? Y is continuous with a constant subgroup sample size = 1 Control Charts: Individual & moving Range (IMR) Stat>Control Charts>Variables Charts for Individuals>I-MR Is my Y in control? Y is continuous with a subgroup sample size >9 Control Charts: X Bar Standard Deviation Stat>Control Charts>Variables Charts for Subgroups>Xbar-S Is my Y in control? Y is continuous with a subgroup sample size between 3 and 5 Control Charts: X Bar Range Stat>Control Charts>Variables Charts for Subgroups>Xbar-R Beauchamp Consulting – Proprietary and Confident
- Slides: 24