Information in this document is subject to change without notice and does not represent a commitment on part of Umetrics. The software, which includes information contained in any databases, described in this document is furnished under license agreement or nondisclosure agreement and may be used or copied only in accordance with the terms of the agreement. It is against the law to copy the software except as specifically allowed in the license or nondisclosure agreement. No part of this user guide may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, for any purpose, without the express written permission of Umetrics. Umetrics patents and trade marks: OPLSO2PLS®, SWE98022296, USA6754543, OSC®, SWE00005637, USA6853923 PLSTREE®, Pending SEQUENTIAL MODELING, Pending DESIGN SPACE, Pending MODDE®, RED MUP®, VALUE FROM DATA®, OPLS®, O2PLS®, O2PLS®, OPLSDA®, O2PLSDA™ PLSTREE™, SPLOT™, EZinfo®, SBOL®, FABSTAT®, BATCH FINGERPRINT®, SIMCA®. ID 2038 MKS Umetrics AB Stortorget 21 SE211 34 Malmö Sweden Phone:+46 (0)40 664 2580 Email:infoumetrics.com Welcome
User Guide to MODDE By Umetrics Version 9 © 1992-2009 MKS Umetrics AB, all rights reserved Information in this document is subject to change without notice and does not represent a commitment on part of Umetrics. The software, which includes information contained in any databases, described in this document is furnished under license agreement or nondisclosure agreement and may be used or copied only in accordance with the terms of the agreement. It is against the law to copy the software except as specifically allowed in the license or nondisclosure agreement. No part of this user guide may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, for any purpose, without the express written permission of Umetrics. Umetrics patents and trade marks: OPLS/O2PLS®, SWE-9802229-6, USA-6754543, OSC®, SWE-0000563-7, USA-6853923 PLS-TREE®, Pending SEQUENTIAL MODELING, Pending DESIGN SPACE, Pending MODDE®, RED - MUP®, VALUE FROM DATA®, OPLS®, O2PLS®, O2- PLS®, OPLS-DA®, O2PLS-DA™ PLS-TREE™, S-PLOT™, EZinfo®, SBOL®, FABSTAT®, BATCH FINGERPRINT®, SIMCA®. ID # 2038 MKS Umetrics AB Stortorget 21 SE-211 34 Malmö Sweden Phone: +46 (0)40 664 2580 Email: info@umetrics.com Welcome Welcome to the user guide for MODDE 9. This is your guide to MODDE and its capabilities. Assumed prior knowledge We assume that you are already familiar with Windows. You should be familiar with the topics discussed in your Microsoft Windows User's Guide, including how to use the mouse select, click, shift-click, control-click, press, drag, and choose from a menu. You should also know how to search through directories to find files. Content This user guide is divided into 21 chapters. Chapter 1 gives a short introduction of how to use MODDE. Chapter 2 presents an introduction to experimental design. Chapter 3 is an overview of the program. Chapters 4 - 20 provide step-by-step procedures for creating and using experimental designs with MODDE. Chapter 21 describes how the Help function works. Appendix A, "Statistical notes" presents short explanations of statistical methods used by MODDE. Appendix B, “Designs” presents short descriptions of the designs available in MODDE. Appendix C: "Optimizer" describes the optimizer feature and the properties of the different optimizer objectives. Appendix D: "Design Space" describes the design space estimation feature. References are available on the references page. User guide edition Tuesday, September 22, 2009. i Table of Contents How to get started with MODDE 1 Installation 1 Starting MODDE 1 Experimental cycle 1 Design phase 2 Defining factors 2 Defining responses 2 Defining objective 2 Analysis phase 2 Explore the data (Worksheet menu) 2 Evaluate the design 3 Fit 3 Review the fit using plots and lists 3 Diagnostics 3 Interpret the model 3 Refine the model 4 Prediction phase (using the model) 4 Introduction to MODDE and experimental design 5 General description 5 What is modeling and experimental design? 5 Objectives of modeling and experimental design 5 Screening models and designs 5 Number of factors in screening designs 6 Number of factors with split objective 6 Response surface modeling (RSM) designs 6 Number of factors in RSM designs 6 Number of factors in Split objective 7 Fit methods 7 Multiple Linear Regression (MLR) 7 Partial Least Squares (PLS) 7 Results 8 Analysis phase 9 Review the model fit 9 Assess model adequacy 9 Prediction - using the fitted model 9 Convention 10 Limitations in investigation names 10 Limitations in factor and response names 10 Case sensitivity 10 Menu reference syntax 10 Suggestions for further reading on experimental designs 10 Overview 11 Application icon and symbol 11 User Guide to MODDE ii Investigation 11 Managing investigations 11 Compatibility with older MODDE versions 12 Organization 12 Command menu bar 13 File menu 13 Edit menu 14 View menu 14 Design menu 14 Worksheet menu 15 Analysis menu 15 Prediction menu 15 Show menu 16 Window menu 16 Toolbars 16 Plots and lists 16 File 17 Introduction 17 New 17 Traditional designs 18 Advanced designs 18 Import design from file 21 Open 21 Save and save as 21 Delete investigation 21 Save plot/list as 21 Revert 22 Send by E-Mail 22 Generate HTML report 22 Complement design 23 Available complementing methods 24 Fold over 24 Estimate squares of selected factors in factorial designs 24 D-Optimal 25 Complement Doehlert 27 Screening to RSM Rechtschaffner 28 Plackett Burman Super-Saturated to Plackett Burman 28 Protect investigation 29 Encrypt/password protect an investigation 29 Lock investigation 29 Print Setup 30 Print Preview 30 Print 30 Print format for plots 30 Exit 30 Edit 31 Introduction 31 Undo 31 Cut, copy, paste, delete, select all 31 Table of Contents iii Add factor, add response, add experiment, insert rows 32 Sort 32 Sorting the candidate set 33 Model / reference mixture 34 Model 34 Reference mixture 36 Generators 37 Edit the model and/or generators before creating the worksheet 38 View 39 Introduction 39 Toolbars 39 Standard toolbar 39 Spreadsheet toolbar 41 Plot toolbar 41 Model toolbar 44 Window toolbar 44 Status bar 44 Dockable windows 44 Analysis advisor 45 Audit trail 45 Favorites 45 Output 50 Notes 50 Add to Favorites 51 Add to Report 51 Full Screen 51 Changing the default options using Investigation Options 52 Plot labels 52 Number format 52 Audit trail 53 Alpha level 54 Coefficients 55 Blocking 56 Confidence level 56 List presentation 57 R2 in plots 57 Residuals 58 Select Factor 58 General Options 58 General page 59 List Options 60 Restore 61 Customize 61 Customize toolbars 61 Customize commands 61 Customize options 62 Customize options 62 Design 63 Introduction 63 User Guide to MODDE iv Factors 63 Responses 63 Constraints 63 Inclusions 64 Objective 64 D-Optimal 64 Design wizard 64 Factors 65 Introduction 65 Factor definition dialog 65 Factor name 66 Abbreviation 66 Units 66 General page 66 Advanced 69 Factor definition spreadsheet 72 Printing the factor spreadsheet 72 Factor manipulations in short 73 Opening the factor definition spreadsheet 73 Adding a factor 73 Modifying a factor 73 Updating the worksheet 73 Copying a factor 73 Deleting a factor 74 Responses 75 Introduction 75 Response definition dialog 75 Response name 76 Abbreviation 76 Units 76 Selecting type of response 76 Limits 76 Regular responses 77 Transformation 77 MLR scaling 77 PLS scaling 77 Derived responses 78 Defining derived responses 79 Modifying a derived response 80 Copying or deleting a derived response 80 Using sets of variables in derived responses 80 Syntax for derived responses 81 Operators and functions in derived responses 81 Qualitative factors in derived responses 82 Linked responses 82 Response definition spreadsheet 83 Printing the response definition spreadsheet 83 Response manipulations in short 84 Opening the responses spreadsheet 84 Table of Contents v Adding a response 84 Modifying a response 84 Deleting a response 84 Copying and pasting a response 84 Constraints and inclusions 85 Constraints 85 Specifying constraints 85 Constraints supported 86 Defining constraints in the spreadsheet 86 Defining a constraint graphically 87 Modifying a constraint graphically 88 Candidate set with a constraint 89 Constraints in qualitative or quantitative multilevel factors 89 Inclusions 90 Inclusions vs. complement design 90 Inclusions added to the worksheet 90 Inclusions as part of the design 91 Generating inclusions 92 Modifying inclusions 92 Objective, model and design 93 Introduction 93 Selecting the objective 94 Screening objective 94 RSM objective 94 Split Objective 94 Paste Data 95 Selecting model and design 95 Designs in MODDE 95 Recommended designs 96 Runs in design 96 Model 96 Screening models 96 RSM models 96 Split models 96 Pseudo resolution for blocked designs 97 Design runs 97 Center points 98 Replicates 98 Total runs 98 Settings 99 Description 101 Blocks 102 D-Optimal 103 What are D-Optimal designs? 103 When do I use D-Optimal designs? 103 D-Optimal pages in the design wizard 104 Design generation criteria section 104 Design alternatives section 106 Candidate set section 107 User Guide to MODDE vi D-Optimal results 108 D-Optimal on the Design menu 109 Generate 109 Candidate set 109 Evaluate 110 Onion plot, onion 3D scatter 110 Design plot 110 D-Optimal onion design 111 Candidate set 111 Generating the design 112 D-Optimal on the Design menu with Onion 113 Evaluate 114 Onion plots 115 Design Wizard 117 Introduction 117 Defining factors 117 Defining responses 119 Selecting the objective 120 Selecting model and design 121 D-Optimal pages in the design wizard 122 Worksheet 123 Introduction 123 Worksheet menu 123 Opening the worksheet spreadsheet 123 Setting run order 124 Curvature diagnostic plot 125 2D and 3D scatter plots 126 Histogram 127 Descriptive statistics 128 Correlation 129 Replicate plot 130 Worksheet spreadsheet 131 Description of the worksheet 131 Missing values in the worksheet 131 Deleting the worksheet 131 Displaying the worksheet 131 Adding experiments in the worksheet 132 Sorting the worksheet 132 Colors in the worksheet 132 Analysis 133 Introduction 133 Organization of the Analysis menu 133 Evaluate 134 Condition number 134 Runs, terms and degrees of freedom 134 Analysis Wizard 135 Open the Analysis Wizard 135 Analysis Wizard content 136 [...]... delivery letter 3 After completing the installation, MODDE needs to be activated with a license file Activation is done either (a) from an internal license server or (b) from Umetrics See the delivery letter, sent to the license administrator at your company, for instructions Starting MODDE Start MODDE by double-clicking its icon • To read about the MODDE software look in Help (contains the same information... create plots and list to review the model and fit and use the Prediction menu to create prediction plots and lists Compatibility with older MODDE versions All investigations from MODDE 3.0 and upwards can be opened in MODDE 9.0 (the reverse is not true) Organization The MODDE window consists of a command menu bar and toolbars 12 Overview Command menu bar With the command menu bar you define or edit your... by MODDE with the extension *.mip When you start a new investigation you select its name and destination directory You can open, save, save as, and delete investigations You can generate the fold over of investigations with fractional factorial designs or Plackett Burman You can double-click a MODDE investigation (a *.mip file), in Microsoft Explorer, to open that investigation 11 User Guide to MODDE. .. 286 Result statistics for k2 288 References 289 Index 291 xii How to get started with MODDE Installation You can install MODDE under Windows Vista, XP, and 2000 Note: You must have administrative privileges to be able to install the software To install and activate MODDE follow the steps described below: 1 Download the password protected installation file from the Umetrics web... Predictions you find the Contour plots, Prediction plots and list, Sweet Spot plot, Optimizer and the Design Space feature For more see the Prediction chapter 4 Introduction to MODDE and experimental design General description MODDE - (MODeling and DEsign) is a Windows program for the generation and evaluation of statistical experimental designs Methods of statistical experimental designs have evolved... causes substantial changes (effects) in the response when it varies 5 User Guide to MODDE In the screening stage one uses simple models (linear or linear with interactions), and experimental designs that allows the identification of the factors with the largest effects in the fewest possible number of experimental runs MODDE supports: Full Factorial, Fractional factorial, L-designs, Plackett Burman,... Fractional factorial, L-designs, Plackett Burman, Rechtschaffner, Onion, and D-optimal designs for screening experiments With mixture factors, MODDE supports the classical axial design when the region is a simplex Number of factors in screening designs Process factors: MODDE supports up to 32 factors in problems involving process factors only Mixture factors: Up to 20 mixture factors are allowed in problems... influence the response; get a map of the system • Make predictions, optimize or find a region of operability MODDE supports: Three-level full factorial, central composite, (CCC and CCF), Box Behnken, Rechtschaffner, Doehlert, Onion, and D-optimal designs for RSM investigations With mixture factors, MODDE supports the modified simplex centroid when the experimental region is a simplex For investigations... used to estimate the coefficients of the model The model represents the relationship between the response Y and the factors X1, X2, etc MODDE uses multiple linear regression (MLR) or Partial Least Squares (PLS) to estimate the coefficients of the terms in the model MODDE recommends PLS when the investigation has a high condition number Multiple Linear Regression (MLR) With Multiple Linear Regression... to T 7 User Guide to MODDE This analysis has the following two objectives: 1 To well approximate the X and Y 2 To maximize the correlation between X and Y in the projected space (between u and t) The dimensionality, number of significant PLS components, is determined by cross validation (CV), where PRESS (Predictive Residual Sum of Squares) is computed for each model dimension MODDE selects automatically . with MODDE Installation You can install MODDE under Windows Vista, XP, and 2000. Note: You must have administrative privileges to be able to install the software. To install and activate MODDE. license administrator at your company, for instructions. Starting MODDE Start MODDE by double-clicking its icon. • To read about the MODDE software look in Help (contains the same information as. Tuesday, September 22, 2009. i Table of Contents How to get started with MODDE 1 Installation 1 Starting MODDE 1 Experimental cycle 1 Design phase 2 Defining factors 2 Defining responses