| Date/Time |
Munchen I
|
Munchen II |
Berlin |
Athen - recorded |
|
Wed
10:30-Noon
|
EU 2018 411
Formula Editor Deep Dive
Mike Muhlada and Audrey Schull
This introductory demo will use examples to explain Formula Editor mechanics, to investigate and fix common errors, and to solidify best practices and useful ‘tricks’ for formula creation. A detailed walkthrough will cover various aspects of the formula builder, including using drag and drop, keyboard shortcuts, and function buttons to create formulas. As we work the examples, we will discuss useful functions, uncover and correct common mistakes, work with subscripts, and more. Best practice topics to be covered include: custom formats, using formulas for data ‘cleanup’, column transforms and using intermediate columns to build more complex formulas. Special emphasis will be placed on JMP 14 new features including User-defined functions, result preview, ‘live’ formula error/warning detection, and customizable function list.
Data Access and Manipulation
Level 1
|
EU 2018 414
What did they mean by that? The Essentials of Text Explorer in JMP
Don McCormack/Bill Worley
People make comments, machines output status. There are massive amounts of words and phrases that could be used to better understand comments or summaries. The question many companies are asking is what to do with these mountains of untapped unstructured data. Text Explorer in JMP offers you a data mining tool to help analyze and better understand your customers, or why that new piece of equipment that was supposed to save your business is not behaving as expected. The presenters will show off the new platform features in JMP 14 along with features from the outstanding features found in the original Text Explorer in JMP 13. Features that quickly and easily help you determine frequently used terms and phrases that are driving consumer comments or machine reliability. Several case studies will be used to show how amazingly easy it is to clean your data. Options like Local Recode or adding stop words and phrases that have little or no meaning to your analysis. Word clouds are popping up all over and you will learn how to generate an interactive word cloud in just a couple of clicks.
The presenters will demonstrate advanced text exploration tools in JMP Pro and how to use Latent Sematic SVD and Topic Analysis within Text Explorer to uncover new insights from your unstructured data. Finally, the presenters will demonstrate that by taking advantage of your transformed unstructured data along with your structured data you can build more predictive and informative models.
Data Exploration
Level 2
|
EU 2018 413
Statistical Thinking for Industrial Problem-Solving
Mia Stephens
Ian Cox
Learn how to exploit data and the visual and interactive nature of JMP to solve your most challenging problems. In this breakout session we introduce the principle of statistical thinking and present a practical, data-driven approach to real-world, industrial problem-solving. We foster the understanding of core techniques and modern statistical methods through examples and simulations, and present an end-to-end case study to unify concepts and broaden applicability.
Data Exploration
Level 2
|
EU 2018 412
Sharing JMP Graphs and Reports
Bryan Fricke and Michael Hecht
JMP is the tool of choice for visually exploring and making discoveries about your data. However, since not everyone is a JMP user, you may need to export your graphs and reports in a form tailored to your audience. In this session, we will demonstrate how to get graphs and reports out of JMP and into forms suitable for sharing. In particular, we will cover exporting graphs and reports in forms ready for presentations, printed journals or articles, and interactive web pages.
Data Visualization
Level 1
|
|
Wed
1:00-2:30
|
EU 2018 418
Simulation Methods in JMP and JMP Pro, With Special Emphasis on the Fractional-Random-Weight Bootstrap
Bill Meeker and
Chris Gotwalt
Simulation-based statistical methods allow us to do some incredible things with data. In this tutorial we will go through a variety of examples that illustrate how JMP has made it easy to take advantage of these powerful methods. We will begin with reliability applications and show how to obtain more refined inferences in situations where there are very few observations using Bayesian techniques in JMP’s Life Distribution Platform and in the Fit Life By X platform. We will then introduce the One-Click Bootstrap in JMP Pro and demonstrate its use on a variety of examples, with special emphasis on the tremendously useful, yet very much underappreciated fractional-random-weight bootstrap. The classic bootstrap, based on resampling, has, for several decades, been a widely used method for computing trustworthy confidence intervals for applications where no exact method is available and when sample sizes are not large enough to be able to rely on easy-to-compute large-sample approximate methods, such a Wald (normal-approximation) confidence intervals. There are, however many applications where the resampling bootstrap method cannot be used. These include situations where the data are heavily censored, logistic regression when the “success” response is a rare event or where there is limited mixing of successes and failures across the explanatory variable(s), and designed experiments where the number of parameters is close to the number of observations. The thing that these three situations have in common is that there may be a substantial proportion of the resamples where is not possible to estimate all of the parameters in the model. The fractional-random-weight bootstrap method can be used to avoid these problems and provide trustworthy confidence intervals, and is only a click away in most of the platforms in JMP Pro. Next we will demonstrate how to do power calculations using the Simulate Feature in JMP Pro. This makes it straightforward to do power calculations for many analyses throughout JMP, particularly designed experiments where the outcome is binary. As a final demonstration of JMP Pro’s simulation capabilities we will show how to obtain confidence intervals on neural network predictions using the Save Bagged Estimates option in the Profiler.
Predictive Modeling
Level 4
|
EU 2018 416
JMP Projects: A New Way to Organize Your Work in JMP 14 - Dan Valente and Aaron Andersen
Many JMP users have experienced an interactive JMP session that left them with many windows and analyses reports opened all at once. While using JMP in this "floating window mode" may work for many, there are situations where having a tabbed interface is preferred. Users also may want to associate certain files with an analysis project and not have the full list afforded by the Home Window. For these reasons, we created the JMP project in JMP 14. Projects provide a single document interface to JMP, a tabbed, re-configurable work space, a place to bookmark files and a window list that lets the user easily navigate various open windows, data tables, scripts and more as well as launch supporting files like .pdfs and .ppt documents. Projects can also be used to quickly open and close many files associated with one or more individual analysis activities, and can be used to keep parallel projects separate where in the past you may have needed to run multiple JMP sessions. Finally, projects in JMP 14 can also give you an easy way to share and collaborate on work with their archiving functionality.
Data Access and Manipulation
Level 1
|
EU 2018 417
New DOE Features in JMP14
Ryan Lekivetz
Bradley Jones
JMP 14 has several new DOE features.
Perhaps the most important is a new design tool for creating Balanced Incomplete Block Designs (BIBDs). These are designs having one treatment factor and one blocking factor. Their special properties are that each treatment level appears the same number of times overall and treatment level appears together with every other treatment level in some block the same number of times. This creates a special and desirable kind of balance. We will provide more details about this kind of design as well as demonstrating how to use the UI of the new tool.
The Custom Designer has added a pair of optimality criteria for generating A-optimal designs and weighted A-optimal designs. An A-optimal design minimize the average variance of the parameter estimates. The A-criterion is similar to the well-known D-criterion in that it focusses on the precise estimation of model parameters. There are two benefits arguing for the routine use of A-optimal designs. First, it is easier to understand how the A-criterion produces good designs compared to the D-criterion. That would make for easier adoption of routine use of optimal designs for practitioners who are new to DOE. Second, weighted A-optimal designs allow for putting different amounts of emphasis on different groups of parameters. For example, one might want to emphasize pure quadratic effects over two-factor interactions and both of these over main effect estimation. We will show how to create both A-optimal and weighted A-optimal designs using the Custom Design tool.
The Compare Designs tool can now accommodate up to 5 designs in the UI and up to 10 designs through scripting. Previously, the limit was three designs. We will demonstrate this additional capability with a UI example and a scripting example including the new A-optimal and weighted A-optimal designs.
Of course, there are many other improvements of the DOE tools for JMP 14. Accelerated Life Test design is much improved. The Fit Definitive Screening platform now allows for considering model terms that do not exhibit heredity. The Choice Design and MaxDiff Design tools both have the ability to simulate responses. This will make it possible to determine whether a given prospective design will be able to reliably detect effects of a specified magnitude. We will be available throughout the conference to demonstrate these new capabilities. Unfortunately, time does not permit including these in the tutorial.
A-Optimal Designs
BIBDs
DOE
Level 2
|
EU 2018 415
Using JSL to Develop Efficient, Robust Applications
Joseph Morgan
This tutorial will focus on JSL constructs that can prove invaluable when developing efficient, robust, production-ready applications. These include expression handling functions, xpath support, namespaces and others. The ideas will be illustrated with examples that the instructor has collected over the past 10 years as he has worked with a wide cross section of JSL users.
JSL Application Development
Level 3
|
|
Wed
2:45-4:15
|
EU 2018 419
Building Dashboards and Applications with JMP
Dan Schikore
Saving an analysis so that it can be repeated on the same or different data tables is an important step in creating an analysis workflow that is efficient and repeatable. The JMP Scripting Language (JSL) can be used along with the drag-and-drop tools available in the JMP Dashboard Builder and Application Builder to quickly build and deploy dashboard and applications. This talk will cover everything from simple dashboards created with “Combine Windows” through full applications with multiple windows and JSL scripts. Learn more about designing reports with data filters, parameterizing applications for run-time selection of columns to analyze, and options for sharing of results, including interactive HTML. The new custom template interface will be demonstrated, which makes it easier and faster to create multiple dashboards or applications with a common look and feel.
Data Visualization
Level 2
|
EU 2018 422
Integrated Process Improvement Using the Second-Generation Quality Tools in JMP®
Laura Lancaster
Level 2
The second-generation quality tools in JMP – Control Chart Builder, Process Capability, Measurement Systems Analysis and Process Screening – were designed with an integrative philosophy to make quality analysis easier and more effective. For example, the new Process Capability platform was designed to reflect the type of control chart used in the statistical process control program, and a capability report was added inside of the Control Chart Builder. Similarly, the Shift Detection Profiler in the Measurement Systems Analysis platform allows quality engineers to make informed decisions about how to design their control chart methodology, taking into account their measurement system so they are alerted to process changes as quickly as possible. Understanding how your measurement system, statistical process control program and process capability assessments fit together is key to improving and maintaining quality. The software’s unique design philosophy makes this simple and straightforward. Additionally, the ability to quickly screen large numbers of processes for stability with the new Process Screening platform will save time, reduce workload and improve quality. This tutorial will use a case study to show how to use each of these platforms with an integrated process improvement approach.
Quality and Reliability
Level 2
|
EU 2018 421
One stop model building with the Generalized Regression platform
Clay Barker
Variable selection, the process of choosing a set of factors or predictors to model a response, is an essential part of the model building process. Luckily, it does not have to be a painful process. The Generalized Regression platform (Genreg) in JMP Pro provides a variety of automated variable selection techniques that make this process fast and easy. After using the platform, we are left with a parsimonious model that will predict well on new data. We will look at using Genreg to build models in a variety of settings: starting with an orthogonal designed experiment and moving to large observational data sets. We will also preview some of the new Genreg features in version 14, like modeling multi-level categorical responses and the Dantzig Selector.
Predictive Modeling
Level 2
|
EU 2018 420
Getting to the Fun Part: How to Prepare your Data for Analysis
Brian Corcoran, Mandy Chambers and Kelci Miclaus
JMP visualization and statistical analysis platforms allow you to discover meaningful signals, trends and insights to your data. When a project requires working with multiple complex data files, getting to that “aha” moment requires necessary data querying, joining, import, exploration and cleaning. It is not uncommon for up to 80% of your time to be spent solely on data preparation in complex or large data analytic tasks. This session highlights the use of several new features in JMP 13 and JMP 14 to get you to that analysis ready state quickly and easily. We will discuss how data can be queried and imported from a database, opened and joined with the JMP Query Builder, and multiple file import options new in JMP 14. JMP 13 introduced virtual joins to JMP data tables, which allows you to specify common keys to link multiple tables. New capabilities in JMP 14 direct a table to listen or dispatch row state changes among linked tables. We will demonstrate how these new features allow virtually joined tables to “talk” to each other by sharing row states for easy data exploration and immediate analysis/dashboard creation without table manipulation. Finally we will tackle the dirty job of cleaning your data, highlighting new features of Recode to demonstrate how we can now replace strings using regular expressions, remove unwanted punctuation, save Recode scripts, reuse and append to them, and even use value labels if the column has them already assigned. These topics will streamline every JMP user's experience with import, table explorations, and data cleaning; leaving more time for fun: data analysis.
Data Access and Manipulation
Level 1
|
|
Wed
4:30-6:00
|
EU 2018 426
Reliability Analysis and Engineering Using JMP
Peng Liu, JMP Principal Research Statistician Developer, SAS
Leo Wright, JMP Principal Product Manager, SAS
Michael Crotty, JMP Senior Statistical Writer, SAS
The state-of-the-art suite of tools in JMP for studying reliability is designed to enable organizations to provide superior products -- a requirement for success in today’s arena of complex products and services. For many years, we have incorporated cutting-edge concepts that integrate robust statistical analysis with dynamic data visualization to help users immediately assess performance, spot trends and diagnose improvement opportunities, even in huge data sets. JMP has recognized that reliability, also known as “quality over time,” is crucial across the board, from consumer and industrial products to health care to defense systems. JMP reliability tools are designed to assist engineers, researchers, and managers to understand reliability data and system reliabilities easier and faster, whether the system studied is relatively simple or thoroughly complex. JMP reliability tools, backed by the powerful JMP infrastructure, are also extensible and customizable. In this tutorial, we will cover the following JMP platforms: Life Distribution, Reliability Forecasting, Degradation, and Repairable Systems Simulation. The platforms are among the essential components of reliability analysis and engineering, addressing problems for both repairable and non-repairable systems. The tutorial will start with an introduction to the censored data and will outline different objectives of reliability analysis and engineering. The instructors will address individual objectives by using examples included with JMP software.
Quality and Reliability
Level 3
|
EU 2018 424
10+ Things You Don’t Know About JMP
Brady Brady
As you already know, JMP is an amazingly rich tool for data analysis and exploration. That also means that you know that there's more than one way to do almost anything in JMP.
The 10 (or more!) tips and tricks shared here will make you more efficient and introduce you to techniques, keyboard shortcuts, formulas, and more.
Come in and level-up from novice to expert, or expert to guru.
Data Exploration
Level 1
|
EU 2018 425
The Multivariate Flavors of JMP: From Continuous to Categorical to Discrete to Functional
Laura Castro-Schilo and Chris Gotwalt
Multivariate data analysis has become an essential skill as the amount of data have skyrocketed and as companies become more data-driven in their decision making. In this extended breakout session we showcase JMP’s unique visual and interactive approach to multivariate data analysis using a variety of approaches that rely on continuous, categorical, discrete, and functional data. We will begin with a foundational discussion of the essence of multivariate analysis: the idea that information contained in large numbers of variables can often be efficiently represented with a smaller number of variables. We then give an overview of JMP’s general tools for analyzing multivariate data. The session will be given in two parts. The introduction will use data from sensory analysis and consumer research to motivate and demonstrate classic unsupervised multivariate methods such as Principal Components Analysis, Multiple Correspondence Analysis, and a new technique in JMP 14, Multiple Factor Analysis. The second part will cover supervised methods for multivariate data using Partial Least Squares and Functional Principal Components. To illustrate these supervised methods we will build models to predict the yield of a batch manufacturing process with data from multiple signal streams. Throughout, we will give clear guidelines as to when each analytical technique is appropriate and will highlight the most important and useful software options.
Data Exploration
Level 2
|
EU 2018 423
New and Advanced Views in JMP 14 Graph Builder
Xan Gregg and Scott Wise
This Breakout Session combines the popular “What’s New in Graph Builder” and “Pictures from the Gallery Advanced Graph Builder Views” sessions.
First, we’ll take a guided tour of several new and refined Graph Builder features in JMP 14. A new interval drop zone makes custom error bars easier. Several kinds of jitter have been added, included a new default. A completely new bar chart style called “packed bars” extends Pareto charts. More element types support parallel coordinates for compact multivariate views. Contours can be smoothed. Elementary statistics are accessible.
Then we will explore how to build popular and captivating advanced graph views using JMP Graph Builder. We have created brand-new pictures for the Gallery 3 journal that features additional views available in the latest versions of JMP. We will feature several popular industry graph formats that you may not have known could be easily built within JMP. Views such as incorporating Bullet Charts, Two-Way Error Bars, Volcano Plots and more will be included that can help breathe life into your analytics and provide a compelling platform to help manage up your results!
Data Visualization
Level 1
|
|
Thu
10:45-11:15
(30 minutes)
|
EU 2018 105
Christoph Gross
SCHOTT AG
Iterating over file sets
'Iterating over file sets' is a common task in the process of script-based data preparation and analysis. Typical scenarios include the automated conversion of non-JMP files (e.g. measurement files, which are coming in on a day-to-day base) and the extraction & collection of intermediary results from file sets, which are to large for a simple join. Rather than talking about what is done with the individual files, I will focus on the iterating framework and will present a set of helpful functions.
|
EU 2018 113
Elie Maricau
BASF Antwerpen
Application of Data Analytics to improve the yield of a chemical production process: learnings and best practices
This presentation discusses an example use case in a chemical production plant where the objective is to improve the product yield. As part of the project, 5 years of historical data (including lab data, sensor data and logbook text data) have been analyzed. The results of the analysis trigger dedicated experiments in the plant, both to validate data-driven findings and to further explore unknown relationships. During the presentation, I will highlight some critical success factors in the project approach (i.e. LSS DMAIC) and I will demonstrate how JMP was used to find relationships hidden in the data. I will demonstrate how JMP can be very effective to quickly aggregate datasets, clean the data, separate noise from actual process variations and to visualize and quantify relationships. Techniques that will be discussed and demonstrated include table operations, data filtering, modeling and DOE.
|
EU 2018 116
Martin Owen
Insight by Design
Project, Insight, Action!!!
As a lecturer and trainer, I am very interested in the capabilities offered by the new Project functionality in JMP14EA. I see this platform extending our abilities to enhance organisational memory, bringing material (datasets, scripts, model and visualisation reports) together from different sources. We can map onto existing file structures and create a new organisational structure via a single document interface. In this presentation I will give examples of the potential for JMP14 Project as a teaching tool. The aim is to accelerate learning and help teams or organizations adopt new work flows, retain insights from prior studies, make better decisions and take action. The flexible Project workspace brings to life compelling stories. It is now very easy to:
- create a workflow of sequential operations, from designing experiments to analysing and documenting key information and actions
- show how information builds over a series of Experimental design work packages
- compare and select from different modelling and visualisation approaches
- create and archive standard reports which update when new data is entered
An added bonus is that it is now easier to manage and curate data as I work - closing a current session ready to regenerate the next day couldn’t be simpler.
|
EU 2018 104
Piet Hoogkamer
Abbott Healthcare Products
Multivariate testing of the similarity of dissolution curves
Dissolution testing of pharmaceutical products is important as it is a surrogate measure of in-vivo dissolution. In vivo dissolution affects the bio-availability, which may affect pharmaco kinetics (blood levels) and as a result may affect safety and efficacy. In case a change is proposed with respect to the manufacturing process, the manufacturing site, or the test method, ‘equivalence’ needs to be demonstrated to obtain a bio-waiver. Similarity of curves may be tested using the mathematical f2 metric, where a value between 50 and 100 suggests similarity. The f2 metric will be explained in detail. However, in case the variability is too high (between tested units, within a timepoint), a multivariate distance based statistic needs to be used instead. Some options will be shown for model dependent and model independent approaches. Use of JMP, both with standard functionality and with a dedicated script, will be illustrated for the Mahalanobis distance.
|
|
Thu
11:30 - 12:00
(30 minutes)
|
EU 2018 115
Paolo Nencioni
A.Menarini M.L. & S. S.r.L.
Using Fit Model platform to predict, in a short time, the shelf life of some formulations, during the development phases of a drug product.
In general, International Council of Harmonization (ICH) guidelines require that a drug product should be evaluated under predefined storage conditions that test its thermal stability. The storage conditions and the lengths of studies chosen should be sufficient to cover storage, shipment, and use. The long-term testing should cover a minimum of twelve months’ duration at the time of submission, while the intermediate and accelerated condition should cover a minimum of six months. During development phases, more than one formulation, can results to meet the project requirements. How can we choose the best one? Surely, we would like to choose the formulation that assure the longer shelf life as possible. Performing a formal ICH stability study give useful responses not before six months. Performing accelerated stability studies using higher temperature than ICH conditions, has permitted to evaluate the behaviour of some formulation, in a relative short time (not more than thirty days). With JMP Fit Model platform, it has been possible to estimate degradation rate in function of the stress factor, arriving to a data-driven decision about the more stable formula. The estimated degradation rates have been used to predict the quality attribute values at time points and storage conditions required form ICH guideline. The predicted results have been compared with actual value of one batch followed using ICH conditions, in order to verify the goodness of the models. The same approach has been used evaluating together the effect of temperature and oxygen level. The models built with JMP has been used to identify real effect of the oxygen content, on the critical quality attributes of the finished product.
|
EU 2018 121
Jacqueline Asscher
Kinneret College on the Sea of Galilee
Industry training session based on a case study: Using JMP to learn how to design split plot experiments
When experiments have a split plot structure, this should be taken into consideration in both design and analysis. A common example is an experiment conducted to improve a process that runs in batches, where some factors are changed between batches and other factors are changed within batches, resulting in runs that are not fully independent. The split plot structure may be inherent in the process being investigated or in the experimental set up, or may be adopted by choice either to save time or money, or to improve the design. In the case study, some factors are related to material preparation, and other factors are related to the processing of the material into the final product. In the training session for practitioners, the JMP tools for designing, evaluating and comparing experiments are used to explore alternative strategies for choosing a design. The candidate designs considered include the default and user specified designs built by JMP and designs built from scratch by the user. For example, it is usually cheaper to include minimal whole plot replication, and we explore the consequences of adopting this strategy. The pedagogical aim is to simultaneously teach the topic and the use of the JMP tools.
|
EU 2018 102
Anne SAINT-AMANS
Limagrain
An add-in to analyze field trial data: the Custom ANOVA
Limagrain, fourth largest seed company worldwide, creates every year new varieties for vegetable and field crops. These new varieties are tested and compared to commercial references based on several variables (yield, length of fruits…). Analysis of variance and mixed models are run to take into account undesired environmental effects (border, ventilation, soil fertility…) and extract the genetic potential of the varieties. JMP possesses a lot of features to realize the required statistical studies. An add-in called “Custom ANOVA” has been developed by Limagrain to gather them. The objective is to offer a complete and interactive report specifically adapted to the group’s researchers needs. A minimalist and intuitive interface enables to select a few options before launching the analyses. The final report contains different tabs, each of them associated with a particular step: descriptive statistics, statistics of the model, validation of the hypotheses, effects study, genetic potential calculation, field maps and multiple comparisons. In addition, the add-in exploits the “expression object” of JSL to provide the ability of saving the script of the report in the data table (like any other JMP report). Thus the user can keep a trace of its results and easily regenerate the whole report.
|
EU 2018 124
Gary Finka, Sonya Godbert, Doug Marsh, Director, Statistics Leader, Senior Scientist
Glaxo Smith Kline
Exploring Functional Data
There are many ways that you can explore how the Xs relate to the Ys when you have a rectangle of data, with columns of variables for each row (samples, runs, batches). But lots of other data that you might want to use doesn’t easily fit into this form. A common example is where, for each sample, run or batch, you have some measure that has been recorded over time. What if you want to relate the shape of this “functional data” to other variables? The challenge is to extract a small number of “features” that capture the useful information from the complete functional data from each “function ID” (the samples, runs or batches). The features can then be used as Xs or Ys in a standard data rectangle. Other challenges include the fact that the data may not be aligned across function IDs: the frequency of collection may be different, there may be an offset, or there might be missing data at some points for some of the function IDs. With examples from a Biopharmaceutical setting we will show how valuable insights can be extracted from this type of data using Functional Data Explorer in JMP 14.
|
|
Thu
2:00 - 2:45
(45 minutes)
|
EU 2018 202
ALFREDO LOPEZ NAVARRO
Telefónica Germany GmbH & Co. OHG
TITANIC Revisited - All Hands on Deck !
We would like to provide a compelling and interesting description of some -not very well known- facts about the disaster of RMS TITANIC, sunk in 1912 after colliding with an iceberg. We all know the history, or so we think. Thanks to Telefónica’s Discovery Methodology —powered by JMP-, we will embark on an expedition diving deep into a see of data points in order to rescue insights from oblivion. We’ll scrap data using the "Internet Open" feature. Join, concatenate and update tables until creating an enhanced database capable of providing plenty of insights. Then, by means of ANOVAs, Categorical and Visual analysis we’ll tell a fascinating data-driven story. Expect a last minute surprise  Of course it’s going to be all hands on deck: Everything is documented and explained so to be uploaded onto the JMP User Community. We would like to open a conversation for other to build on top of this. JMP on board with us!
|
EU 2018 201
Philip Ramsey
University of New Hampshire and North Haven Group
Model Validation Strategies for Designed Experiments Using Bootstrapping Techniques with Applications to Bio-pharmaceuticals
There are two different goals to statistical modeling: Explanation, and Prediction. Explanatory models often predict poorly (Shmueli, 2010). Often analyses of designed experiments (DOE) are explanatory, yet the experimental goals are prediction. DOE a best practice for product and process development where one predicts future performance. Predictive modeling requires partitioning the data into training and validation sets where the validation set is used to assess predictive models. Most DOEs have insufficient observations to form a validation set precluding direct assessment of prediction performance. We demonstrate a “Balanced Auto-Validation” technique using the original data to create two copies of that data; one a training set and the other a validation set; the sets differ in row weightings. The weights are Dirichlet distributed and “Balanced”; observations contributing more to the training data contribute less to the validation set (and vice versa). The approach only requires copying the data and creating a formula column for weights. The technique is general allowing one to apply predictive modeling techniques to smaller datasets common to laboratory and manufacturing studies. Two bio-pharma process development case studies are used for demonstration. Both cases have large validation sets combined with Definitive Screening Designs. JMP is used to demonstrate analyses.
|
EU 2018 212
robert Reul
Isometric Solutions
Using JMP to Eliminate Discrimination in the Workpalce
Eliminating discrimination in the workplace Possibly the one goal shared by applicants and employers is the ability to seek and secure employment on a full set of non-discriminatory characteristics: knowledge, skill, and experience. The unfortunate reality is that these are all too often trumped by other more easily discernable factors that have very little to do with the job performance, and worse, embody unethical and illegal practices when hiring applicants for a job. Can data analytics be used to exclude discriminatory bias during the hiring decision? By leveraging extensive academic literature on personality and performance, several frameworks emerge that serve as outcome vectors for predictive models namely the Big-5 personalities and Trustworthiness, Likeability, and Confidence. But what to do about the predictors? Through experimentation with facial image processing data from sources such as Google and Microsoft showed predictive promise. By using a series of analytic methods that screen for predictive potential and reduce dimensional complexity, predictive prowess emerges with a completely non- discriminatory set of latent variables that inform the “whom to interview” / “whom to hire” recommendations based purely on cultural personality fit. This presentation will reveal the challenges and successes of this effort proving it’s not only feasible – it may become the new normal.
|
EU 2018 210
Paul Nelson and Andrew Macpherson
GSK
Definitive Augmentation of Definitive Screening Designs
Definitive Screening Designs (DSDs) uniquely address the key needs of many experimenters. How else can we explain the rapid and enthusiastic adoption of DSDs since their discovery was published in 2011? 13- or 17-run DSDs for 5-7 factors are go-to designs for many experimenters when screening for the few driving factors. Along with ‘Fit Definitive Screening’ in JMP, you potentially have a simple, efficient and effective experimental workflow to find the important main effects, interactions and curvilinear behaviours of these factors. If only 3 of the factors are active you can fit the full second-order RSM model and achieve screening and optimisation in one step. But what if more than 3 factors are active? When ambiguity occurs is there a simple next step? Or does the complexity of this situation become a barrier to adoption of DSDs? In this presentation, you will first hear about real-life examples that demonstrate the value of DSDs. You will then see there are simple ways to augment these DSDs, ensuring that the structure and properties can be preserved, so the benefits of DSDs are maintained. Consequently, more people in more situations can benefit through the workflow of sequential DoE and DSDs.
|
|
Thu
3:00 - 3:45
(45 Minutes)
|
EU 2018 211
Rob Lievense
Perrigo
Incorporating QbD Elements to Minimize Spray Pattern Variation of an Active Drug Spray Coating Process Using JMP Image Analysis and the DOE Platform
Obtaining acceptable dosage uniformity, for pharmaceutical tablets coated with an active drug, is dependent upon mitigating variability among the dosage units. A much higher level of precision is required from the drum tablet coating process to apply active coatings than non-functional coatings. The spray system is a key process element that is very challenging to optimize due to the need for specialized equipment to obtain volumetric measurements. An alternate approach involves trials that spray the pattern onto a simple, black paper substrate and obtain digital images of the spray pattern. The image analyzer is used to create a table of intensity values, which quantify the pattern. Graph builder plots of the intensity profile clearly illustrate the variation in spray between and among multiple spray guns working as an efficient analog to the typical volumetric measures. The profiles are summarized into outputs that are used with the custom DOE platform to optimize the spray system process. JMP is utilized to incorporate Quality by Design (QbD) elements to define the design space. Optimization of this space greatly enhances the process so exacting standards set for dosage uniformity can be robustly met.
|
EU 2018 208
Olivier Brack
Statistique Industrielle Khi2 Consulting
Design of Experiment in Numerical Simulation
On the eve of the publication of a new AFNOR standard (FDX06 081) devoted to DOE on simulation and industrial tests, we propose to present an example of JMP application to solve an optimization problem performance of a tracked vehicle undercarriage, based on numerical simulations Following the processing of data from unplanned simulations to address the problem, new simulations were performed to obtain more relevant information. These new simulations were also carried out with a view to obtaining data enabling to compare DOEs, with a pedagogical objective in the framework of training at the Ecole Polytechnique. This example shows the utility, too often neglected, of the DOEs in numerical simulation. From the design of the tests, to the recalibration of the theoretical model on real experimental data, passing, obviously by the statistical treatment (notably taking into account the experimental variability, too often and so unfairly put aside in simulation) and graphical of the results of the plans, this presentation highlights JMP's performance in achieving the requirements of DOEs in numerical simulation.
|
EU 2018 205
Thomas Zelikman
NNE
Predictive control synthesis of an industrial control loop with Jmp
An industrial setup is presented, where Jmp was used to import and join files from from several different temporal data sources including vibration measurements, machine operation data and video files. With JSL in Jmp, it was possible to create a script that performs all necessary conversions and preconditioning of data such as extraction of sound from video files, spectral analysis of sound and vibration and synchronization of temporal data with geometrical measurement positions on the specimens produced in the process to enable linking process events with measurements on the output. Multivariate methods were used to create a prediction of the output suitable for exporting predictor expression to the controller. Design of Experiments methods were used to identify how the output should be affected to close the control loop and obtain the desired response.
|
EU 2018 204
Per Vase
NNE
How JMP 14 can predict performance of future batches. A game changer in validation of pharmaceutical processes.
Many companies within the pharmaceutical industry still pass a Process Performance Qualification (PPQ) simply on the fact that the estimated Ppk on three consecutive batches are better than the acceptance criteria. However, validation is about predicting the future – NOT the past. So even though the three batches may have a sufficient Ppk this is not sufficient. The prediction for future Ppk values also needs to look promising. International Society of Pharmaceutical Engineering (ISPE) has published a discussion paper “Evaluation of Impact of Statistical Tools on PPQ Outcomes” on how this prediction can be made from variance component calculation of between and within batch variance converted to process capability indices. One of the reasons for this method still not being widespread used has been the lack of statistical software packages with this calculation method as a standard. However, it is now in JMP 14 in the 3-Way chart where Between and Within Sigma Capability is being calculated. Examples where this method has been used for PPQ evaluation will be shown in JMP 14 both with successful and unsuccessful evaluations. It will be compared with the traditional method mentioned previously.
|
