JMP 7 and Minitab 15 Thomas A Little
JMP 7 and Minitab 15 Thomas A. Little Ph. D. 07/07/07 1 © TLC, SS 0 070402
Audience Description This presentation is designed for those individuals who are interested in understanding the differences in the design, function and capabilities of JMP 7 versus Minitab 15. Particular attention is made to those features and functions used for Six Sigma/Lean project application. Software JMP 7 and Minitab 15. Limitations 762 North 470 East American Fork, UT 84003 1 -925 -285 -1847 drlittle@dr-tom. com www. dr-tom. com This presentation is limited to those features and functions of greatest interest to users in the scientific, business, engineering and six sigma/lean communities. An attempt was made to review the features and functions in both applications from a user’s perspective. TLC actively consults with both applications and finds features and functions in both applications that are best in class. Any disagreements about observations found in this presentation should be addressed to the author who welcomes opposing points of view. 2 © TLC, SS 0 070402
Presentation Outline Section I General Interface and Ease of Use Section II Lean Six Sigma Activities Define Measure Analyze Improve Control Section III Extended Capabilities Section IV New Features and Conclusions 3 © TLC, SS 0 070402
JMP Version 7. 0 Overview Power § JMP provides more analytical tools, graphs, depth, scripting and features that are used to solve real world problems § Static and dynamic visualization of data via meaningful graphs and options. Version 7 added significantly to this capability. § JMP is particularly good at large data sets and multivariate modeling § JMP benefits from SAS’s core capabilities and years of development § JMP version 7 improves linkage and data transfer to SAS Speed § Single define, multiple output § All graphs and reports in the same window, powerful table commands not available in excel § Control, command function to manipulate them all Ease of Use § JMP organization simplifies the windows, text and graphs integrated § Simplified interface to complex activities such as Fit Y by X and Fit Model § Ease of data and table manipulation. 4 © TLC, SS 0 070402
Minitab Version 15 § Both Minitab (MT) and JMP are far superior for data analysis than using Excel § MT is a mature, full featured product with years of user input and product features § MT was selected by GE and Honeywell as the early six sigma engine of choice when JMP was just developing version 4. At the time they were correct, MT was the better, more mature product. The world has spun since that time and JMP has surpassed MT’s capabilities in all three of the areas of greatest interest to users; speed, power and ease of use. § MT release 15 remains a blessing and a curse. Blessing due to its years of application development and familiar tools. Curse due to its old, awkward interface and software design. § MT continues to be a much slower application once the data sets rises above 100, 000 observations. 5 © TLC, SS 0 070402
Section I General Interface and Ease of Use General design Windows Organization 6 © TLC, SS 0 070402
General Design, Tables Minitab uses projects and worksheets as major file formats; where projects are collections of worksheets. JMP has similar capabilities. Table commands for Minitab and JMP are very similar and JMP has some additional table features not found in MT. More table manipulation tools in JMP and more readable file formats. Advantage JMP 7 © TLC, SS 0 070402
Data Table Size Opening and Manipulating Large Data Sets* File Size (rows) Time to File Open Time to Display One Histogram JMP Minitab 1 M <1 sec. 13 sec. 1 sec. 90 sec. 5 M 5 sec. 15 sec. 6 sec. 100 sec. 20 M 24 sec. Failed. 35 sec. Failed to display Minitab failed to load 20 M rows, all 3 columns, only one column loaded. Advantage JMP takes seconds and Minitab takes minutes to manipulate data. If datasets are large as they are in many transactional environments MT is not a tenable solution. Even with moderately sized data tables MT feels slow on response times. *MT JMP evaluation PC used was running Vista, 1. 80 GHz Duo, 2 GB RAM 8 © TLC, SS 0 070402
Data Tables and Graphs Linked In MT there is row identification capability; however, no real connection between the graph and table. JMP makes the connection which allows for ease of row location, data and graph manipulation. Major Advantage JMP 9 © TLC, SS 0 070402
Menus MT displays the analysis method by name. JMP layers the analysis based on one variable, two, paried and multiple Xs and multiple Ys. Menu Pros and Cons Analysis of One Two Paired or Many variables of any data type. Minitab is easier to use if you are looking for a specific type of analysis by name. JMP’s Analyze tools are organized based on single, two, paired and multiple factors. JMP is generalized and easier to learn and remember. This is particularly true of Green Belt level training. Major Advantage JMP 10 © TLC, SS 0 070402
Graphs and Analysis File: Clean. Minitab uses a separate graph and session window for most of the output. This feature is very annoying in Minitab and slows down the user and the time to analysis understanding. It is a very old school design. JMP keeps all reports and graphs together in one place. Advantage JMP 11 © TLC, SS 0 070402
Subsets JMP is visual and intuitive when creating subsets. MT does it with formulas, row numbers or brushing. Advantage JMP 12 © TLC, SS 0 070402
Formulas and Functions JMP has a complete and rich set of integrated functions for data and string manipulation. MT has fewer overall functions and they are spread out and segmented in the Calc function. Advantage JMP 13 © TLC, SS 0 070402
Section II Six Sigma Activities Define Link to process flow analysis Measure Process capability and MSA Analyze Hypothesis testing and performance modeling Improve Design of Experiments and Robust Tolerance Design Control SPC 14 © TLC, SS 0 070402
Define, Process Flow Analysis Minitab and JMP are developing partnerships for linking process mapping, value stream mapping and Lean manufacturing analysis tools into their respective analytical engines. i. Grafx for example has both JMP and MT connections. Advantage - Draw 15 © TLC, SS 0 070402
Process Capability, Minitab Normal File: Cn MT’s process potential study is poorly named in this graph. Missing PPM and sigma quality. 16 © TLC, SS 0 070402
Process Capability, JMP Normal JMP’s six graph analysis is hard to find without training; however, it is very good and is easy to interact with. It is a feature under control charts. JMP includes sigma quality in its report and has more secondary options. It allows for nonnormal fit selection on the fly. Advantage JMP’s second capability graph is poorly named. It should be called process potential. 17 © TLC, SS 0 070402
Nonnormal Capability Fitting File: Skewed JMP and MT have similar fitting capabilities, JMP has an interactive interface and an overall better report. Advantage JMP 18 © TLC, SS 0 070402
Nonnormal Capability in Minitab MT is missing the sigma quality level and the quantile plot to look at the quality of the fit. The sixpack report is a better option in general when using MT. 19 © TLC, SS 0 070402
Minitab Pareto MT does not allow for easy selection of comparison groups and does not allow for DPU summary tables from the Pareto platform. Cannot directly generate a cost or severity weighted Pareto plot. 20 © TLC, SS 0 070402
JMP Pareto JMP allows for easy grouping variables, DPU summary tables and cost and severity weighted Pareto generation. Advantage JMP 21 © TLC, SS 0 070402
Surface Plots, MT Both MT and JMP have nice surface characterization capabilities. MT is slow to generate and difficult to manipulate. Control over the image is slower and has less options. 22 © TLC, SS 0 070402
Surface Plots, JMP 3 D visualization in JMP is excellent in either the contour or surface plots. JMP allows for up to 100 gradients and MT allows for only 11 in the contour plot. JMP’s Surface Profiler is based on Open GL a full 3 D graphics engine. Advantage JMP 23 © TLC, SS 0 070402
GR&R in MT File: Gage study ANOVA analysis is similar, JMP has the variability graph which is better at displaying variation patterns. MT removes some of the misleading AIAG reports and provides an easier to read report format. MT is missing the secondary breakdown of variation. 24 © TLC, SS 0 070402
JMP GR&R Functionality JMP has the variability chart that is better for showing variation patterns in the data; however, it is missing the control chart for outlier detection and the summary graphs. JMP needs to add the control chart, summary graphs and secondary breakdown of the variation patterns to be best in class. 25 © TLC, SS 0 070402
Bias and Linearity, MT The linearity graph in MT is in error. The reference line should be relative to the mean and not to zero. MT does not have the secondary breakdown of bias by part and by comparison group. MT does have the p-values for all of the comparisons which is very desirable. 26 © TLC, SS 0 070402
Bias and Linearity, JMP’s reports are correct and more detailed in general. JMP is missing the p-values for the bias errors. JMP displays the impact to the standard deviation based on rotation effects. Advantage JMP 27 © TLC, SS 0 070402
Attribute GR&R, MT MT has a very good and very detailed agreement analysis report; however, it is poor on graphing and labeling of effectiveness. Agreement/effectiveness by part, prob(miss), prob(false alarm), bias report and escape rate are all missing in MT. 28 © TLC, SS 0 070402
Attribute GR&R, JMP’s attribute GR&R report is very good and covers agreement and effectiveness very well. It is missing bias and escape rate. JMP’s graphs are better at showing agreement (blue line) and effectiveness (red line). Advantage JMP 29 © TLC, SS 0 070402
Context Sensitive Fit Y by X This is where JMP shines over Minitab and provides the user with the proper analysis depending on the data type. JMP automatically switches between four different analytical platforms depending on the column attributes. Advantage JMP 30 © TLC, SS 0 070402
Correlation Fit Y by X Correlation studies, exploratory data analysis, fit special, group by, etc. , this is where JMP outperforms MT on option after option. Advantage JMP File: Factory RSM 31 © TLC, SS 0 070402
Fit Y by X Contingency Tables JMP and MT have similar summary table capabilities; however, MT is missing the visualization graphs. Advantage JMP File: Failures 32 © TLC, SS 0 070402
Multiple Regression, N-Way, ANCOVA MT requires detailed statistical and modeling training to remember the names of all of the types of ANOVA. Once the analysis is preformed there is not an easy to use suite of tools and secondary graphs for the user to interact with for further visualization, characterization and optimization. Tools are segmented and not well integrated for optimization. File: cement 33 © TLC, SS 0 070402
Multiple Regression, N-Way, ANCOVA Simple model definition no matter the data type. In addition to the detailed statistical summary tables JMP offers a full suite of graphs for visualization, characterization and optimization. Advantage JMP File: cement 34 © TLC, SS 0 070402
Design of Experiments - Design DOE in Minitab is awkward to use for designing experiments as it does not allow for the direct design of the experiment in line with the problem that needs characterization. Minitab uses a candidate points method for customization and augmentation. This is very old school and tedious for the user. Covariates are not part of the design, they are secondary in the analysis. Minitab does not allow for correct factor identification when designing the experiment. There are many more factor types than those allowed by MT. MT fails the ease of use test for DOE. File: Yield 35 © TLC, SS 0 070402
DOE Analysis, MT Analysis flow MT’s analysis tools for DOE are segmented, do not flow well and the optimizer is missing a more intuitive set of controls for constraining, fixing, optimizing and predicting the response. MT’s DOE design and analysis flow is segmented, complicated, not seamlessly integrated and has too many steps. 36 © TLC, SS 0 070402
Design of Experiments in JMP custom designs match the problem. Any combination of factors, factor types, covariates, blocking sizes, categorical factors and mixtures with a minimum sample size. Simple to define the model terms to be characterized. Allows the most flexible environment for DOE treating the engineer and scientist as the customer. JMP is best is class for DOE. JMP wins on DOE ease of use. In JMP the DOE design always fits the problem. 37 © TLC, SS 0 070402
DOE Analysis in JMP is the Same Fit Model Engine In JMP learn one set of tools and use them for a variety of characterization, DOE, modeling problem solving activities. JMP’s profiler allows for improved visualization and control of the transfer functions. Major Advantage JMP 38 © TLC, SS 0 070402
JMP’s Simulator Linked to Transfer Functions Optimize performance, improve robustness and predict full distribution at target. MT does not have this capability. Set and evaluate tolerances. Major Advantage JMP 39 © TLC, SS 0 070402
Power and Sample Size JMP has sample size calculation for counts per unit and for estimating the standard deviation. MT identifies sample size for replicates for two specific forms of DOE and JMP does not. JMP also has a sigma quality converter and calculator. Minor Advantage JMP 40 © TLC, SS 0 070402
SPC JMP 6 to Minitab 14 Comparison 11/22/2005 SPC Control Charts for Subgroups Xbar R Xbar S Presummarize Delta to Target, subgroup Z subgroup Control Charts for Individuals Run Chart I/MR Z/MR individual Delta to Target, individual Levey Jennings Control Charts for Small Mean Shifts UWMA (moving average) EWMA CUSUM Control Charts for Attributes P NP C U Multivariable Control Charts T 2 Multivariate EWMA JMP 6. 0 MT 14. 1 Y Y Y N N Y Y Y Y Y N N Y Y © 11/22/05 MT and JMP’s capabilities are quite similar. MT offers more charts; however, JMP’s charts are easier to manipulate and are better for larger data sets. JMP needs to add the short run Z and delta to target charts. Both platforms allow for phased control charts to show before and after effects. Advantage - Draw 41 © TLC, SS 0 070402
Section III Extended Capabilities Reliability Multivariate Time Series Graphs Advanced Modeling Summary 42 © TLC, SS 0 070402
Reliability MT offers reliability planning tools for sample size determination and JMP does not. JMP has stronger modeling and multivariate tools for reliability modeling. Advantage - Draw 43 © TLC, SS 0 070402
Multivariate JMP has a richer set of tools for multivariate analysis. Factor analysis and principle components analysis are in the multivariate platform and are harder to locate from the menu. Advantage JMP 44 © TLC, SS 0 070402
Time Series JMP and Minitab similar tools and capabilities. JMP has a few more options and the ease of use and graphical manipulation makes it superior to MT. Minor Advantage JMP 45 © TLC, SS 0 070402
Graphs JMP offers similar graphs to MT; however, it outperforms in the profiler, contour profiler, surface plot and custom profiler options. MT does not have the same rich tools for optimization and robust design. Advantage JMP 46 © TLC, SS 0 070402
Advanced Modeling Tools JMP offers a much richer and versatile set of modeling tools and analytical methods. Neural nets, recursive partitions and nonlinear modeling are all available modeling tools in JMP. Advantage JMP 47 © TLC, SS 0 070402
For A More Detailed Comparison For a more detailed comparison of JMP versus MT take a look at the JMP 6 to MT 14 comparison table. 48 © TLC, SS 0 070402
Summary § JMP is in general a superior product § JMP is world class for regression, modeling, DOE, and simple studies such as process capability and MSA and the user interface is very well designed § JMP is easier to use, more powerful, much faster in completing analysis of data and needs to address some of the minor gaps identified in this comparison § Having two great applications is good for the market and keeps both applications improving to meet customer needs and expectations § MT is a good application and has a rich set of tools. JMP is a great application and has an overall better designed and better integrated tool set. § Helping companies understand why Excel is not enough for analysis is the greatest opportunity § Minitab must address the ease of use, some missing tools and speed issues. 49 © TLC, SS 0 070402
762 North 470 East American Fork, UT 84003 925 -285 -1847 drlittle@dr-tom. com www. dr-tom. com 50 © TLC, SS 0 070402
- Slides: 50