-
Discovery Summit 2018
SAS World Headquarters, Cary, NC
23-26 October 2018
Abstracts
-
Give Us Your Bottlenecks, Your Problems, Your Hidden Frustrations
John Sall, Co-Founder and Executive Vice President, SAS
We like listening to customers. We even like hearing about the difficulties, bottlenecks and frustrations that users encounter in doing real-world work. That's because we need to know how users are approaching modern problems and where gaps exist. It's that feedback that allows us to know to focus our time, talent and ingenuity on the next iterations of JMP. Simply put: We listen, we create, and we fix. I think you'll enjoy some of the creations and fixes in JMP 14, which became available earlier this year. In my plenary talk, I'll show you the things I think you'll be most excited about.
-
How Charts Lie — and How They Make Us Smarter
Alberto Cairo, Professor of Professional Practice, Knight Chair in Visual Journalism, University of Miami
Enthusiasm about data visualization has exploded in the past decade, with professionals from varied disciplines adopting it as a tool for discovery and communication. This is great news, as visualization is indeed a language that is as powerful as it is flexible. However, many views about how we should design and interpret visualizations are still naive. Many journalists, designers and analysts think, for instance, that a data map or graph alone can “prove” a statement or opinion, or that visualizations are always intuitive, univocal and unambiguous. This talk debunks those myths and many others surrounding visualization.
-
Quiet: How to Harness the Strengths of Introverts to Transform How We Work, Lead and Innovate
Susan Cain, Chief Revolutionary, Quiet Revolution
Did you know that introverted leaders often deliver better results than extroverts? That the most spectacularly creative people tend to be introverts? That the most innovative thinking happens alone and not in teams? One of the central challenges of any business is to bring out the best in its employees. Yet when it comes to introverts – who make up a third to a half of the workforce – our leadership strategy mainly consists of asking them to act like extroverts. This is a serious waste of talent and energy. In an enlightening, relatable and practical talk, Susan Cain shows us that introverts think and work in ways that are crucial to the survival of today’s organizations. How can you structure your organization so that the best ideas dominate, rather than those of the most vocal and assertive people? How do introverts’ and extroverts’ different personalities cause them to solve problems and evaluate risk differently? What do introverts know about creativity that the rest of us should learn? Drawing on her original research and the latest in neuroscience and psychology, Cain will radically change your view of the best way to develop leaders, manage teams, make smart hires and stimulate innovation.
-
20 Tips and Tricks to Make JMP® Work Better for You Regardless of Experience Level (US 2018 137)
Michael Anderson, Senior Engineer, Corning
- Topic: Data Access and Manipulation
- Level: 1
JMP is incredibly powerful, and each new version adds many useful features, but also changes or removes options and moves things around. For new users it can be daunting, and for longtime users it can result in missing out on efficiency-boosting workflows. This paper will step through 20 pieces of advice – from simple tips to more involved tricks – that will help you get the most from JMP with the least amount of work, so you can focus on your data rather than the tools. These tips will help you with bringing data into JMP, putting together your analysis and getting your results presentation-ready!
-
A Good Query Can Set You Up for Success (US 2018 117)
Brian Corcoran, JMP Director of Research and Development, SAS
- Topic: Data Access and Manipulation
- Level: 1
Data preparation can be a time-consuming, complicated process that takes valuable time away from conducting an analysis. Sometimes, though, a good query not only can simplify data preparation, but can create insights of its own. This talk will highlight a JMP Query Builder example, using World Bank data easily obtained from the web. An exploratory query will be created, with the goal of making it easily reproducible and shareable.
-
All Models Are Wrong, But Simulation Helps Identify the Best of the Bunch (US 2018 145)
Robert Anderson, JMP Senior Statistical Consultant, SAS
- Topic: Predictive Modeling
- Level: 2
Correctly identifying the best possible model and determining which factors are genuinely important are always vitally important tasks, but never easy. Holdback validation is often used to suppress overfitting and avoid including non-genuine terms in a model. However, it is not a foolproof method, especially when working with small data sets. The model you obtain is often dependent on how the training and validation rows are assigned. A single validation column cannot be relied on to point to the “best” model. However, by using many different validation columns, a clearer picture starts to emerge. Using the Simulate function in JMP Pro and some simulated data sets, this presentation will demonstrate how refitting models using multiple validation columns allows the most frequently occurring and most likely model to be identified. It will also demonstrate that this approach works even for data sets with as few as 30 rows.
-
Applying Multivariate Statistical Techniques in the Study of Chocolate and Its Potential Effects on Cardiovascular and Neurovascular Disease (US 2018 215)
Mason Chen, Stanford University OHS Program
Charles Chen, Lean Six Sigma Master Black Belt and Industry Consultant
- Topic: Predictive Modeling
- Level: 2
Many people like eating chocolate, but may have concerns about health risks, especially regarding cardiovascular or neurovascular diseases. Using JMP 14, multivariate statistical techniques were applied to define a health biometric to help with choosing a healthy chocolate for patients with heart disease. Chocolate, made from cocoa beans, contains flavonoids. Flavonoids are the most abundant polyphenols in the human diet and have antioxidant properties that can prevent aging. Flavonoids are also beneficial for heart disease and diabetes patients, as is a diet that is low in saturated fat, trans fat, sodium and cholesterol, and high in dietary fiber. Cocoa flavanols, a class of flavonoid, promote healthy blood flow from head to toe. The heart, brain and muscles depend on a healthy circulatory system. Data were collected on more than 20 chocolate ingredients from over 60 different types of chocolate. A multivariate correlation study has found a strong negative correlation between cocoa and sugar, and a strong positive correlation between dietary fiber and iron. Most dark chocolate contains more cocoa and less sugar. Dietary fiber and iron are highly correlated because of the high cocoa concentration. The above two correlations can be further explained by conducting a hierarchical clustering analysis to separate the dark, milk and white chocolates by cocoa and calcium factors.
-
Automated and Interactive Analysis of JMP® Crash Data (US 2018 150)
Shannon Conners, JMP Director of Research and Development, SAS
- Topic: Data Access and Manipulation
- Level: 2
The JMP team, like many of our customers, is simultaneously engaged in building new product versions and analyzing problem reports from production releases. Customer bug and crash reports provide precious insights into customers’ experience of JMP product quality. JMP began prompting customers to email Mac crash reports in JMP 12 and added built-in support for emailing Windows crash reports in JMP 13. Technical Support and Development collaborate to identify, investigate and fix unsolved customer crashes. The process starts when a report is emailed to JMP Technical Support and processed in an hourly batch. New crash signatures are loaded into a SQL Server database and screened to identify them as novel or potentially related to a known issue; then details are sent back to the track. I refresh my crash investigation environment frequently with a JSL script that calls JMP Query Builder and table manipulation tools like Split and Summarize, and embeds scripts and web links in tables to enable quick investigation and distribution. Imported and derived data tables open as tabs inside a JMP 14 Project. I use this crash data project to identify additional occurrences of existing crashes and find unsolved crash groupings requiring further investigation by JMP developers.
-
Automated Data Imputation: A Versatile Tool in JMP® Pro 14 for Handling Missing Values (US 2018 130)
Milo Page, JMP Research Statistician Developer, SAS
- Topic: Predictive Modeling
- Level: 2
JMP Pro 14 includes a new Automated Data Imputation (ADI) utility, a versatile, empirically tuned, streaming, missing data imputation method. We recommend it for handling missing data as a pre-processing step to predictive model fitting. It empirically tunes to your data set to extract the underlying structure, even in the presence of missing data. It also respects training and validation partitions and interfaces seamlessly with predictive models. It is developed using powerful matrix completion methods with some added extensions for robustness and flexibility. This talk will focus on when ADI is appropriate and how to use it in JMP Pro 14. I will also outline a recommended workflow for processing data with missing values and demonstrate ADI’s performance on some examples.
-
Basic JSL: Working With Column Dialogs and Buttons (US 2018 224)
Byron Wingerd, JMP Systems Engineer, SAS
- Topic: JSL Application Development
- Level: 1
In JMP, it is easy to capture and reuse scripts. Often the next step is to build custom dialogs, which use these scripts with different tables and column names. These basic projects can easily get derailed by difficulties in capturing file names and paths as well as useable column names. In this talk, I will cover methods for capturing paths, files and column names, as well as an essential JSL programming strategy that will make working with these tricky elements much less treacherous. Whether you've been writing JSL for a while or you are just getting started, these tips and tricks for "peeling function onions" will prevent tears so that you can spend your time developing your applications rather than trying to diagnose cryptic problems.
-
Beyond ROC Curves: Finding Meaning and Balancing Trade-Offs (US 2018 205)
Karen Copeland, Owner, Boulder Statistics
Tarek Zikry, Technical Intern, SAS
Mia Stephens, JMP Academic Ambassador, SAS
- Topic: Predictive Modeling
- Level: 1
For dichotomous classification models (good versus bad, negative versus positive, yes versus no, etc.), an ROC curve and the associated area under the curve (AUC) are often used to describe the ability of the model to differentiate the two groups. While this is a start, the ROC curve can be hard to understand, and the AUC can be misleading outside the context of the modeling problem. For some models, the overall misclassification rate is key; for other models, there is a cost associated with misclassifications. This cost can be very different for a false negative versus a false positive. We will discuss dichotomous models for prediction and their associated ROC curves, lift curves and alternative displays of model performance. We will also discuss methods for selecting optimal cutoffs for classification and measures of performance based on the specified cutoff. To facilitate the selection of the optimal cutoff for classification models and the calculation of performance metrics, we have developed an interactive and graphical JMP application. We will demonstrate this application with a compelling case study.
-
Building a Modeling Culture With Survey Data (US 2018 106)
Scott Reese, Scientist, Procter & Gamble
Amy Phillips, Principal Scientist, Procter & Gamble
- Topic: Data Exploration
- Level: 1
To improve and support P&G’s data-driven decision making, our team has been assessing the quantitative analytical needs of our internal JMP users. We use the analogy that providing internal support is similar to managing a consumer goods product. Instead of a typical product and package that we would market and sell to consumers, we translated our support efforts into a combination of the product (necessary tools), package (delivery of training) and communication (awareness of benefits). As part of this effort, we conducted internal surveys to understand which JMP platforms were being used, the existing skill levels of users and what areas of future learning were desired. This talk focuses on demonstrating basic consumer survey analysis using the Categorical platform with the results of the internal JMP usage survey. We will also discuss how we approached designing and placing the survey to encourage hearing from as many of our users as possible.
-
Building a Set of Statistical Tools With Scripting in JMP® (US 2018 135)
Sean Hardy, Biostatistician, IDEXX Laboratories
- Topic: JSL Application Development
- Level: 2
JMP has many built-in applications for standard statistical analyses that are user-friendly and accessible. Frequently, however, there are cases of repeatable analyses specific to one person, process or industry that might be missing from the core features of JMP and would be desirable to some, but perhaps not all analysts. The JMP Scripting Language (JSL) provides the capability to develop your own tools in the form of add-ins for specific analyses, reference tables and much more. You can even build add-in packages that can be easily shared with other JMP users. The possibilities for the tools themselves are nearly endless, and the only major barrier to building basic tools is understanding how to write scripts in JSL. In this presentation, I will go through the process of designing and building one of these tools. I will provide a brief walk-through of add-in creation for beginners, as well as a couple of examples showcasing how truly creative you can be.
-
Data Exploration Using Interactive HTML Data Filters in JMP® 14 (US 2018 151)
Bryan Fricke, JMP Senior Software Developer, SAS
- Topic: Data Exploration
- Level: 1
Since JMP 11, users have been able to export interactive HTML JMP graphs for data exploration without the aid of JMP. Until JMP 14, data exploration has been limited to informative tool tips and data selections linked between graphs. Now, JMP 14 adds support for data filters, which can be used to visually explore which data categories and/or value ranges wield the most influence over exported graphs. In addition to all the currently supported graph types, treemaps are now supported. This paper demonstrates how interactive HTML graphs can be used to gain insights into the relationships between variables in a completely visual manner. In particular, we will explore relationships between country metrics such as gross domestic product (GDP), economic complexity and World Happiness Report scores.
-
Design and Optimization With Nonlinear Constraints in JMP® (US 2018 108)
Casey Volino, Senior Engineering Associate, Corning
- Topic: Design of Experiments
- Level: 2
This talk will focus on incorporating nonlinear constraints in both the design and analyses of experiments. JMP allows us to constrain the experimental design space with disallowed combinations, but it does not currently obey those constraints for optimization. Working with JMP Profilers, a method will be presented that incorporates nonlinear constraints in optimization problems in the form of disallowed combinations. Multiple examples will be shown that address both design and optimization. All data and the JMP journal used in the presentation will be made freely available.
-
Detecting Small Shifts in Your Process (US 2018 134)
Annie Dudley Zangi, JMP Senior Research Statistician Developer, SAS
Tonya Mauldin, JMP Senior Analytics Software Tester, SAS
- Topic: Quality and Reliability
- Level: 2
Cumulative sum, or CUSUM control charts, are well established as being more efficient in detecting small shifts than Shewhart charts. However, the V-mask in CUSUM charts can be more challenging to interpret than in traditional control charts. In this talk, we will present examples that illustrate how the new CUSUM charts can detect small shifts in the mean that are undetectable in Shewhart charts. We will also preview the new CUSUM Control Chart platform options and compare chart efficiencies with average run lengths.
-
Do Golf Handicaps Always Level the Playing Field? (US 2018 131)
David Trindade, Founder and Owner, Stat-Tech; and Instructor, Santa Clara University
- Topic: Data Exploration
- Level: 2
Golf handicapping is meant to “level the playing field” and allow players of different abilities to compete equally in tournaments. Each golfer has a handicap based on a formula using the 10 best scores in the most recent 20 rounds. In competition, the handicap is subtracted from a golfer’s gross score to determine the net score. The lowest net scores are the winning scores. From a statistical viewpoint, is the handicapping system fair and performing as expected? One analysis approach is to compare graphs of the gross and net scores versus handicaps. Gross scores are expected to increase with handicaps, but net scores should be reasonably comparable across handicaps. For this talk, we use the analytical and visualizing capabilities of JMP to investigate the relationship between gross and net scores versus handicaps for several different tournaments run at a private country club. In the process, we uncover several interesting scoring differences for tournaments among men, women and seniors. We illustrate using JMP software's Fit Y by X, Distributions, Graph Builder, Outlier Screening and By variable formula functions. An “aha!” moment occurs when we discover that the handicapping system for a certain tournament type resulted in outcomes not indicative of a level playing field.
-
Exploring Fundamentals of Acceptance Sampling and Control Chart Monitoring in JMP® (US 2018 110)
Andrew Karl, Statistician, Adsurgo
Heath Rushing, Principal Consultant and Co-Founder, Adsurgo
- Topic: Quality and Reliability
- Level: 1
Recently, a medical device customer wanted a script to evaluate different, standard acceptance sampling plans using operating characteristic (OC) curves. They also wanted to find the limit from an average outgoing quality (AOQ) curve for rectifying inspections. Although they were experienced JMP users, they did not know about the availability of single- and double-sample OC curves found only in the JMP Starter menu. Furthermore, they had not utilized Graph Builder for evaluating common deviations (nothing goes as planned) from these sampling plans or for developing AOQ curves. Although scripting can be quite effective and efficient, writing scripts from scratch can be overwhelming for new JMP users. However, the existing capabilities of JMP can be exploited and easily adapted for routine and/or recurring analysis of common quality questions. This talk with demonstrate the adaptation of the existing capabilities of JMP to explore and evaluate deviations from standard quality approaches such as acceptance sampling, process control and capability, and design of experiments.
-
Happy Classrooms, Happy World (US 2018 230)
Robert Carver, Senior Lecturer, Brandeis University International Business School; and Professor Emeritus, Stonehill College
Volker Kraft, JMP Senior Academic Ambassador, SAS
- Topic: Data Visualization
- Level: 2
Undergraduate courses in statistics continue to grow and evolve, with an emphasis on teaching statistical thinking and engagement throughout the entire process of statistical investigation. The ASA’s 2016 Guidelines for Assessment and Instruction in Statistics Education (GAISE) report challenges university instructors to use abundant digital data and modern software tools to allow novice students to learn experientially, thereby bridging the gap between theory and decision-making with data, fostering intellectual curiosity and exploration of “what if” scenarios, and facilitating the communication of findings and results. It can be challenging to find real data sets that are simultaneously engaging and understandable to undergraduates, as well as suitable to illustrate a thorough statistical investigation. We suggest using data from recent releases of the World Happiness Report to stimulate statistical thinking in an undergraduate first statistics course. The combination of JMP software's intuitive, visual design and the global importance of this data can make for a very happy classroom experience. This talk will simulate a classroom demonstration of exploration, question-generation, and analysis, culminating in presentation-ready visualizations.
-
Importing Web Service Data: The New HTTP Request in JMP® 14 (US 2018 231)
Bryan Boone, JMP Principal Software Developer, SAS
- Topic: JSL Application Development
- Level: 2
Government and private entities are increasingly making their data available over the internet via RESTful web services. JMP 14 has a new feature called HTTP Request, available through JSL, that allows you to access these RESTful services. This presentation will provide several examples on how to use HTTP Request to bring this new data into JMP. There will also be an introduction to what RESTful services are and how JMP uses them.
-
Integrated Process Improvement Using the Second-Generation Quality Tools in JMP® (US 2018 213)
Laura Lancaster, JMP Principal Research Statistician Developer, SAS
Annie Dudley Zangi, JMP Senior Research Statistician Developer, SAS
- Topic: Quality and Reliability
- 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 session will use a case study to show how to use each of these platforms with an integrated process improvement approach.
-
Interactive Profiler as a Supplier Specification (US 2018 107)
Jim Lamar, Quality Manager, Saint-Gobain NorPro
- Topic: Predictive Modeling
- Level: 1
Using JMP, we were able to create a predictive model that solved a problem that had been plaguing the company for decades. The result of over a year’s worth of designed experiments, both within the business and at one of our major suppliers, was two major findings: 1) the quality of their raw material was the ONLY useful factor in predicting our final product quality, and 2) the supplier’s product had two previously unknown raw material quality parameters that contributed to the problem. The interaction of these two parameters was the key to our eventual success. One of these parameters is a feature of our supplier’s raw material. They could control the other parameter with their processing. The traditional raw material specification approach is to define specific limits for each parameter independently. For our needs, the acceptable limits for parameter B depended on the value of parameter A. Breaking from tradition, we met with the supplier and generated an interactive profile that allowed them to input their raw material quality measurement A, and then control the other important factor B to whatever level was needed to ensure that their final product would make an acceptable raw material for us.
-
Introduction to JMP® Text Explorer Platform (US 2018 204)
Jeff Swartzel, Scientist, Procter & Gamble
Tracy Desch, Scientist, Procter & Gamble
Scott Reese, Scientist, Procter & Gamble
- Topic: Data Exploration
- Level: 2
Do you have unstructured text data? Would you like to understand how to use JMP Text Explorer to aid your analysis? Are you intrigued by the possibility of modeling some kind of outcome variable (like an online star rating) using text as the basis for the inputs? JMP Text Explorer is a powerful, hands-on tool that can be used to gain insights from unstructured text data. In this short presentation, we will introduce you to the basics of using the platform. We will cover everything from the initial data curation and analysis steps all the way through more advanced modeling techniques. This session will be highly interactive, with the speakers asking for audience input on curation decisions. Along the way, a few principles will be shared to help approach real-world questions.
-
JEDIS: The JMP® Experimental Design Iterative Solver (US 2018 143)
Jason Sheldon, Research Staff Member, Institute for Defense Analyses
- Topic: Design of Experiments
- Level: 2
The JMP Experimental Design Iterative Solver (JEDIS) is an add-in for JMP software that helps automate the design of experiments process within JMP in a user-friendly manner. JEDIS builds multiple test designs in JMP over user-specified ranges of sample sizes, signal-to-noise ratios (SNR) and alpha (1-confidence) levels. It then automatically calculates the statistical power to detect an effect due to each factor and any specified interactions for each design. When finished, JEDIS presents the statistical power versus design metrics in interactive plots and stores the data in an easy-to-use format. JEDIS supports generating factorial and optimal designs, but does not currently support generating split-plot designs. The JEDIS Light feature can compute power for a premade design table, including premade split-plot designs, over ranges of SNRs and alpha levels. You can download a copy of JEDIS from here.
-
JMP® Into Retirement Using the Reliability Platform in JMP (US 2018 229)
Don Lifke, Systems and Component Reliability Engineer, Sandia National Laboratories
- Topic: Quality and Reliability
- Level: 2
The powerful Reliability platform in JMP is often overlooked and underutilized. This talk demonstrates some of the basics of the Reliability platform by answering the seemingly simple question, “How long will my retirement endure?” Why just guess at this extremely important figure when planning for retirement? Use JMP to explore this question! This was accomplished using Reliability and Survival → Fit Life by X, along with historical data from retirees. Uncertainties of this prediction were also quantified. The optimum retirement age was addressed, considering the fact that retiring earlier draws less income, but for a longer period. In addition, JMP was used to model and more simply visualize the employer’s retirement planning tool to optimize the most financially desirable retirement age. Come join me in exploring the most enjoyable chapter of life – retirement!
-
Just Survive Somehow: An Exploration of the Storylines, Characters and Audience Reactions in Our Favorite Post-Apocalyptic World (US 2018 219)
Ruth Hummel, JMP Academic Ambassador, SAS
Clay Barker, JMP Principal Research Statistician Developer, SAS
- Topic: Data Exploration
- Level: 2
JMP provides a variety of tools that make life easier for a data analyst, from obtaining and manipulating data to visualizing and analyzing it. In this talk, we will show how we used JMP to tackle a very important challenge: answering questions concerning our favorite television show about zombies. Are there trends in viewership or how episodes are received by viewers? Are there “turning point” episodes? What tends to make characters on the show survive longer? We will walk through obtaining this data from the web, preparing the data for analysis, visualizing trends, and finally, analyzing the data in hopes of answering our questions. A variety of analysis platforms in JMP and JMP Pro will be used, including Fit-Y-By-X, Text Explorer and Generalized Regression, and we will consider some common analysis techniques as well as some uncommon ones. After all, we’re just getting started…
-
Machine Learning Makes Predictive Models: JMP® Helps Make Them Explanatory (US 2018 118)
Tony Cooper, Principal Analytical Consultant, SAS
Sam Edgemon, Principal Analytical Consultant, SAS
- Topic: Predictive Modeling
- Level: 2
Machine learning is about creating predictive models. If the target (aka response) is nominal, then a good model will maximize true positives and minimize false positives. Designed experiments will use many of the same analytical algorithms, but the focus is on causality and the appropriate recipe for success. JMP users may be more used to this paradigm, and often look at models and ask, “Why is the prediction so high? How do we improve it? What is important about the business?” These questions are asking predictive models to become explanatory and therefore can be hard to answer; issues such as restricted ranges on the variables and multicollinearity can make it difficult to go from a predictive model to an explanatory model. Further, numeric summaries of models do not encourage subject matter experts to ask questions. The difference between a predictive model and root cause analysis can be overlooked; yet we know correlation does not imply causation. The art of explaining data is getting lost in the push for advanced modeling techniques. The humble Profiler in JMP is a powerful tool for making models talk. This presentation reminds us that, even for a data miner, it is the Profiler and Graph Builder that will make the difference.
-
Model Validation Strategies for Designed Experiments Using Bootstrapping Techniques With Applications to Biopharmaceuticals (US 2018 218)
Chris Gotwalt, JMP Director of Statistical Research and Development, SAS
Philip Ramsey, Principal Lecturer, University of New Hampshire; and Owner, North Haven Group
- Topic: Predictive Modeling
- Level: 2
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 is 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 data sets common to laboratory and manufacturing studies. Two biopharma 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.
-
Modeling Consumer Feedback With Text Mining Features in JMP® Pro (US 2018 109)
Shankang Qu, Statistician, PepsiCo
Mumu Wang, Senior Scientist/Statistician, PepsiCo
- Topic: Predictive Modeling
- Level: 2
This study is intended to evaluate customer reviews and purchasing behavior using supervised learning sentiment analysis. The verbatim (i.e., textual data) collected from consumer feedback and comments on beverage products is classified into categories such as complaint, praise and suggestion. Through the text mining models created with JMP Pro, we demonstrated the feasibility of a hybrid algorithm implementation where hand-built classifiers are combined with empirical learning from the data. An automated system will connect the algorithms to incoming voice-recorded feedback with the cumulated verbatim and classify the document into one of the three categories. The algorithms adapt in response to new data and experiences to improve prediction quality over time. Our service associates manually coded 7,507 documents from consumers through phone, email, social media, chat and e-commerce. In this presentation, we will show how to build text mining models using approaches such as neural networks. In the confusion matrix, we achieved a 6% misclassification rate in training and 11% in validation. Slightly higher rates were obtained by using bootstrap forest and nominal logistic models. Comparing the modeling results and the original coding on the verbatim, we also found some data entry issues. The models have been trained to capture documents that were misclassified by humans.
-
Monitoring Targeted Changes With Cumulative Sums (US 2018 217)
José Ramírez, Chief Statistician, Amgen
Jon Weisz, JMP Senior Vice President, SAS
- Topic: Quality and Reliability
- Level: 2
Shewhart charts are a direct plot of the data and are good at exposing different types of deviations from statistical stability. However, the Shewhart chart is not as sensitive to small changes in the process average; it is in this type of situation where we need a more sensitive chart. Cumulative sum (CUSUM) charts plot the cumulative differences between a process variable and a target, which makes them particularly sensitive to small deviations in the process mean from target. In this talk, we introduce the concepts behind CUSUM charts, and show examples of how easy it is to develop these charts using the new CUSUM Control Chart platform in JMP 14. This new platform emphasizes the use of decision limits rather than the traditional V-Mask, facilitating the development and interpretation of the charts. We will also discuss the use of cumulative score (CuScore) charts to detect changes in the parameters of a given model.
-
New DOE Features in JMP® 14 (US 2018 211)
Bradley Jones, JMP Distinguished Research Fellow, SAS
- Topic: Design of Experiments
- Level: 2
JMP 14 has several new DOE features. Perhaps the most important is a new design tool for creating balanced incomplete block designs (BIBDs). These designs have one treatment factor and one blocking factor. Each treatment level appears the same number of times overall, and each treatment level appears together with every other treatment level in some block the same number of times. This creates a special kind of balance. We will provide more details about this design and demonstrate 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 minimizes the average variance of the parameter estimates. The A-criterion is similar to the well-known D-criterion, in that it focuses on the precise estimation of model parameters. There are two benefits to A-optimal designs. First, it is easier to understand how the A-criterion produces good designs than it is with the D-criterion, making it easier for new DOE practitioners to adopt. 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.
The Compare Designs tool can now accommodate up to five 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 that includes the new A-optimal and weighted A-optimal designs.
-
New in JMP® 14: Balanced Incomplete Block Designs (US 2018 221)
Ryan Lekivetz, JMP Senior Research Statistician Developer, SAS
- Topic: Design of Experiments
- Level: 2
In JMP 14, new to the design of experiments menu is a design tool for creating balanced incomplete block designs. 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 each treatment level appears together with every other treatment level in a 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 demonstrate how to use the UI of the new tool.
-
Optimal Designs With Axial Values (US 2018 203)
Cameron Willden, Statistician, W. L. Gore & Associates
Willis Jensen, Global Statistics Team Leader, W. L. Gore & Associates
- Topic: Design of Experiments
- Level: 2
Optimal designs have become increasingly popular due to their flexibility and broad applicability, especially with robust implementations in software tools such as Custom Design in JMP. However, traditional optimal response surface designs tend to have lower power for quadratic terms due to high collinearity. Central composite designs (CCD) with off-face axial values tend to have significantly less collinearity among quadratic effects, and result in higher power for quadratic terms, better D-efficiency, and lower average prediction variance (i.e., I-optimality) relative to their corresponding equal-sized optimal designs. We propose a new optimal design algorithm and provide an accompanying JMP application to generate designs that combine the relative advantages of CCD and optimal designs into a single design. This modification to the coordinate-exchange algorithm for generating optimal experimental designs incorporates off-face axial runs, and we call the resulting designs “optimal designs with axials” or ODA designs. These ODA designs generally outperform both CCDs and current optimal designs for quadratic response surface models evaluated over cuboidal experimental regions. We show examples of how to easily generate ODA designs using our JMP application and compare them to other competing designs to show their advantages.
-
Pictures From the Gallery 3: Select Advanced Graph Builder Views (US 2018 206)
Scott Wise, JMP Global Enablement & Training Manager, SAS
- Topic: Data Visualization
- Level: 1
If a picture is worth a thousand words, then the visuals that can be created in JMP Graph Builder could be considered fine works of art. This journal presentation features how to build popular and captivating advanced graph views using JMP Graph Builder. Based on the popular Pictures From the Gallery journals, we have created a brand-new Pictures From 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.
-
Sharing Interactive Web Reports in JMP® 14 (US 2018 132)
Josh Markwordt, JMP Software Developer, SAS
John Powell, JMP Principal Software Developer, SAS
- Topic: Data Visualization
- Level: 1
Interactive HTML supports many of the exploratory features of JMP, including point identification, linking, brushing and – new in JMP 14 – data filtering. Saved reports can be shared with colleagues or uploaded to a web server to be viewed in any modern web browser. In addition to support for data filtering, interactive HTML in JMP 14 adds support for the treemap in Graph Builder. This paper demonstrates when and how to use interactive HTML to share and explore JMP results.
-
Supercharge Your User Interfaces in JSL (US 2018 113)
Peter Mroz, Statistical Programmer, Janssen Pharmaceutical
Justin Chilton, JMP Senior Associate Test Engineer, SAS
- Topic: JSL Application Development
- Level: 2
The user interface of an application should be easy to understand and use. Good user interfaces will result in engaged users, fewer frustrations and great user experiences. This talk will focus on how to supercharge your user interfaces using a variety of techniques in JSL. Most people are familiar with using the JMP data table to display and interact with a grid of values. If you delve into JSL, you will discover that you can do similar things by using an object called a Table Box. You can populate a Table Box with the String Col Box, String Col Edit Box, Number Col Box, Number Col Edit Box, Check Box and Radio Box objects. What if you want to display text in different fonts, styles or sizes, or change the foreground or background colors in the Table Box grid? What if you want to display a column of clickable buttons? What if you want to display a column of icons representing the status of a row? Or how about a column of mini-graphs? The Col Box is the answer! The Col Box is a special type of column object that can contain any other display box. Having this ability allows you to improve the user interfaces of your applications. This talk will show numerous examples of how to use the Col Box display object, as well as an implementation in a real-world application that greatly improved usability. In addition, we will show other supercharging techniques, including using icons in buttons and menus for more intuitive actions, using hover-help or tooltips, using tabs to declutter things, and providing search functionality for long picklists to speed selection.
-
Surface Damage Characterization Using Image Analysis, Human Visual Rankings and Customized Algorithms (US 2018 146)
Evan Bittner, Statistical Engineer, Corning
Christine Cecala, Senior Research Scientist, Corning
David Brockway, Senior Statistical Engineer and Distinguished Associate, Corning
Casey Volino, Senior Engineering Associate, Corning
Eric L. Null, Senior Chemical Measurement Scientist, Corning
- Topic: Data Exploration
- Level: 2
Durability of cover glass on displays is an important attribute for consumer devices. To assess cover glass resistance to surface damage, quantitative metrics are needed that can discriminate between different samples in laboratory testing while maintaining a relationship with human visual rankings to accurately simulate user experience. To move from subjective, highly variable rankings to robust quantitative metrics, the VIRTUAL Eye measurement system was developed. Key features include two edge lights to illuminate samples and an overhead camera to capture images of surface damage. With images of surface damage on hand, JMP is used to drive the data processing, quantitative analysis and user feedback. Usable pixel intensity data is extracted from images; in-house algorithms were created and developed to extract the metrics from that data. Established platforms such as Graph Builder and Quality and Process pushed the analysis of data and metric creation forward, while scripting provided power, flexibility and efficiency throughout the development process. The tools in JMP helped provide a path to quantifying the visual perception of surface damage.
-
The Fun Part: My Favorite Things About JMP® 14 (US 2018 214)
Mandy Chambers, JMP Development Tester, SAS
- Topic: Data Visualization
- Level: 2
What are some of your favorite things about data analysis and JMP? Some of my favorite things in JMP 14 include new features as well as some enhanced ones. After this talk, they may become your favorites as well. The new multiple file import capability allows you to load dozens or hundreds of files expediently and concatenate them into a single JMP data table. Column names can now be made shorter when linking virtually joined tables together, making reports and graphs with cleaner titles and axis labels. Data filters are more shareable among linked subsets or by groups, and local data filtering has been expanded, so it is accessible to all the linked subsets. This presentation highlights features that will get you to that analysis-ready state quickly and efficiently.
By importing multiple HR data files (anonymized internal SAS data), a personnel flow matrix can be generated to represent how many people joined, were promoted or left the company, as well as head count per division, performance ratings and new hire longevity. By virtually joining the tables and using the shareable data filter, you can remove or include the specific data that is important for your reporting. These topics will streamline every JMP user's experience with import capabilities, table explorations and data filtering. Adding those features to your favorite things list will leave more time for data analysis – the fun part!
-
The Multivariate Flavors of JMP®: From Continuous to Categorical to Multiple-Source Data (US 2018 207)
Laura Castro-Schilo, JMP Senior Associate Research Statistician Developer, SAS
- Topic: Data Exploration
- Level: 2
Multivariate data analysis has become an essential skill as the amount of data has skyrocketed and companies become more data-driven in their decision making. In this session, we showcase JMP and its unique visual and interactive approach to multivariate data analysis using a variety of approaches that rely on continuous, categorical and multiple-source 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 general tools for analyzing multivariate data in JMP. We 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. Throughout, we will give clear guidelines as to when each analytical technique is appropriate and highlight the most important and useful software options.
-
The Wide Technical Facility of Functional Data Explorer (US 2018 226)
Peter Hersh, JMP Systems Engineer, SAS
- Topic: Data Exploration
- Level: 2
One of the marquee features in JMP Pro 14 is the Functional Data Explorer. But what is "functional data" and how exactly do we "explore" it? Functional data is everywhere. It takes the form of sensor data, transactional data, chemical spectra – the list goes on. The common thread is that it can be challenging to analyze. Moreover, we generally don't want to analyze the functional data directly; we want to work with the underlying information – the functions that are producing the observed data. The Functional Data Explorer (FDE) helps us do this. It serves as a tool for both exploratory analysis and dimension reduction to help us use the functional information in other modeling techniques. In this presentation, Hersh will show the new analytical problems JMP can answer using FDE through several case studies from industrial, chemometric, and the financial domains. Along the way, he will demonstrate FDE and some of the tips and tricks he has learned while helping customers come to understand how powerful this new platform truly is.
-
Time Trials: JMP® Emerges as the Clear Winner Versus Other Analytic Industry Competitors (US 2018 115)
Erin Wright, Informatics Scientist, Perrigo
- Topic: Data Exploration
- Level: 1
The analytics industry has soared in the past few years, and it is beginning to feel like everyone is trying to build their own analytics tool for the market. As an aspiring data scientist, I have had the opportunity to test-drive several tools, and I am here to show you why I will put myself behind the wheel of JMP over the other tools every time. In this paper, I put JMP up against some of the most well-known analytic tools in the industry. One data set, six tools, same graphical goals – software review. See how JMP stacks up against Excel, SPSS, R, Tableau and Power BI when I put each to the test on the famous MTCARS data set.
-
Tumor Response Visualization in Clinical Trial Oncology Studies (US 2018 149)
Kelci Miclaus, JMP Life Sciences R&D Manager, SAS
- Topic: Data Visualization
- Level: 1
Solid tumor oncology studies have become a major clinical trial focus area and present unique challenges in showing that a new therapy actually works. Due to complex study designs, many traditional statistical techniques (like traditional survival analyses) are not well suited to detect early efficacy signals. Three common visualization techniques have now become mainstream in evaluating multiple measured lesions in solid tumor studies. In this presentation, we will illustrate how to use JMP Graph Builder to build Swimmer Plots, Waterfall Plots and Spider Plots for evaluating tumor response in an early-stage clinical trial. New enhancements to Graph Builder in JMP 14 that allow increased flexibility to control the variables contributing to multiple elements will be highlighted.
-
Understanding Wild Lemurs’ Soil Consumption Using Regularized Linear Models in JMP® (US 2018 142)
Jiangeng Huang, PhD Candidate in Statistics, Virginia Tech
Brandon Semel, PhD Candidate in Fish & Wildlife Conservation, Virginia Tech
- Topic: Data Exploration
- Level: 2
In this study, we sought to explain the function of soil consumption in critically endangered wild lemur populations based on nutritional data collected from four groups of diademed sifakas (Propithecus diadema). These groups of lemurs inhabited continuous and fragmented forests in central-eastern Madagascar across five seasons. Biological data in nutritional studies proposes many challenges including multiple variables, multicollinearity between variables and a relatively small number of observations. Advances in regularized linear models, such as elastic net, enabled us to select important features and estimate these effects in one step, addressing the challenge of multicollinearity by combining lasso and ridge regressions at the same time. Here, we use the regularized regression function in JMP to reveal several important variables that help explain the soil consumption behavior of wild lemurs.
-
Using the JMP® Recurrence Analysis Platform to Determine When It Is Better to Replace Rather Than Repair Analytical Equipment (US 2018 144)
Tanya Davis, Senior Biometrician, Zoetis
- Topic: Data Exploration
- Level: 2
Pharmaceutical research and development involves the use of many different types of equipment to measure the critical quality attributes of new drug formulas. As with any equipment that is used regularly, repairs and maintenance are needed to keep instruments operating properly. Recurrence analysis, within the Reliability and Survival menu in JMP, is an effective technique for determining repair rates and trending over time. Recurrence analysis allows for the cost of the associated repair to be tracked; the cost of future repairs is compared with the cost of replacement to determine the best strategy to move forward. Text exploration with word clouds, added to recurrence analyses, enhances the information gained as words and phrases documented in work order verbiage are dynamically linked to recurrence trends over the life of the equipment. The results of the combined analyses allow for data-driven maintenance planning as well as sound justification for instrument replacement.
-
Utilizing JMP® Database Query, Dashboard, and JSL to Expedite the Development of Powerful JMP® Apps (US 2018 202)
James Pappas, Senior Principal Statistician, Medtronic
J. Bruce Dunne, Director, Medtronic
- Topic: Data Visualization
- Level: 2
Each year, for the development of new LigaSure™ devices and generators, Medtronic engineers and scientists use a wide spectrum of data, ranging from supplier component data to final product performance data. Proper data management, timely analyses and data-driven decision making are critical for Medtronic. However, there are constraints and barriers that potentially compromise the ability to do these things effectively. These include, but are not restricted to, the limited availability of dedicated database developers, programmers and data analysts.
The purpose of this work is to utilize the JMP 14 database query and dashboard platforms in concert with JSL and the interactive HTML reports to analyze and visually illustrate device performance. The tools within JMP 13 allow an easy way to create deployable data analysis applications for a variety of internal customers, including high-level dashboards for management and detailed interactive analytical tools for product development engineers. The applications apply across all facets of the business and can be created quickly with minimal programming effort.
JMP 14 demonstrations will include examples of apps currently used to visualize and assess product performance and the relatively simple steps required in creating their respective scripts. Benefits of this are far-reaching, as more engineers become engaged in both data collection and analysis efforts related to their respective processes.
-
Visualization of 3D Parameter Spaces in 2D: A Novel Method for Data Exploration (US 2018 209)
Stefan Nikles, Product Engineer, Analog Devices
- Topic: Data Visualization
- Level: 2
Several JMP platforms exist for visualizing parameter spaces: Contour Plot, Overlay Plot, Surface Plot and Scatterplot 3D, to name a few. These work well when exploring data containing a single dependent variable as a function of two independent variables, but it is more difficult when one needs to explore data as a function of three independent variables (e.g., a three-factor DOE). Users must commonly resort to presenting multiple plots for all values of one or more of the parameters, making it difficult to discern important relationships. In many cases, however, the primary concern is simply knowing where the dependent variable exceeds a specific threshold. For example, under what conditions does the signal-to-noise ratio exceed unity? Or where in the parameter space is the p-value less than 0.05? In this presentation, a method is described for computing a single contour level from multiple plots and overlaying each into a single plot. This method employs a GUI-enabled JSL script to compute the contours, which are then plotted using the Graphics Scripting feature within the Bivariate Platform. The use of this method was essential in validation of a new measurement technique for our devices.
-
What Is My MSA Really Telling Me? (US 2018 129)
Lisa Estey, Senior Staff Scientist, IDEXX Laboratories
Mike Atkinson, Senior Research Scientist, IDEXX Laboratories
- Topic: Quality and Reliability
- Level: 1
Measurement system analysis (MSA) is a statistical tool commonly used in industry by process and quality engineers. Through MSAs, operators gain an understanding of measurement uncertainty as it relates to repeatability, accuracy/bias and the components of variation that may influence test reproducibility. The introduction of Six Sigma methodologies at IDEXX expanded the use of MSAs by product development and problem-solving teams. Here we will examine three gauges posing challenges to the teams and demonstrate the JMP tools used to assess their performance, including Graph Builder, Gauge R & R, EMP and Variance Components. The first MSA assesses a new analytical tool purchased by the R&D organization for the separation of materials by molecular size. Through gauge studies, the researchers were able to identify sources of measurement instability in the system. The other MSAs characterize two analytical tools that were adapted to help troubleshoot manufacturing variation in porous membranes. These MSAs demonstrate common issues observed with destructive tests, and the importance of sample homogeneity when characterizing a new measurement system.
- Beginner: 1
- Intermediate: 2
- Advanced: 3
- Power user: 4
-
Analyzing Direct Versus Digital Optic Systems Using Survival Analysis in JMP® (US 2018 308)
Kevin Eng, Statistician, US Army ARDEC
Douglas Ray, Statistician, US Army ARDEC
-
Deep Web Financial Vulnerabilities: Where Hackers Find Success Accessing Ill-Gotten Financial Gains (US 2018 310)
Abu Bakar Seddeke, Student, Spears School of Business, Oklahoma State University
Preston Lange, Student, Spears School of Business, Oklahoma State University
-
Exploring Female Nobel Laureates in JMP® Featuring HTML5 (US 2018 305)
Hui Di, JMP Senior Test Engineer, SAS
-
Finding a Web Request Timeout Problem (US 2018 307)
Melanie Drake, JMP Principal Systems Developer, SAS
-
Inspection Equipment Prove-Out Using Beta Regression (US 2018 311)
Thorsten Roberts, Statistician, US Army ARDEC
-
JMP® Helps the Army Develop Dispersion Requirements for New Weapon Systems (US 2018 312)
Christopher Drake, Statistician, US Army ARDEC
Kevin Singer, Statistician, US Army ARDEC
-
MetaEval: A JMP® Add-In to Evaluate a Claim Coming From a Meta-Analysis (US 2018 302)
S. Stanley Young, CEO, CGStat
Paul Fogel, Independent Statistical Consultant
-
Metrology Site Reduction Analysis of All Combinations Using NChooseK Matrix JSL Function (US 2018 313)
Daniel Sutton, Statistician, Samsung Austin Semiconductor
-
Optimizing the Biological Control of Insect Pests, Reducing Insecticide Use and Minimizing Economic Injury With JMP® (US 2018 309)
Jane Pierce, Associate Professor, New Mexico State University
-
Predicting Determining Factors of Asian Countries’ Growth Rates for Business Opportunity (US 2018 303)
Christine Manglicmot, Student, Oklahoma State University
Imogen Schramm, Student, Oklahoma State University
-
Predicting Protein Structure From Highly-Correlated Factors Using the Partial Least Squares Platform in JMP® 14 (US 2018 301)
Stan Siranovich, Principal Analyst, Crucial Connection
-
Predictors of Childbirth Satisfaction in the United States Military (US 2018 304)
Melissa Gliner, Senior Health Policy Analyst, US Army Office of the Surgeon General’s Decision Support Center
Kenneth Kovats, Senior Nurse Analyst, US Army MEDCOM Analysis and Evaluation Division
Dawn Garcia, Leader of Women’s Health Analytics, US Army MEDCOM Analysis and Evaluation Division
Richard Thorp, Deputy Chief of Analysis and Evaluation, US Army MEDCOM
-
Tolerance Identification Through RNA Expression Profiling of Renal Allograft Rejection (US 2018 314)
Neal Smith, Pathologist, Massachusetts General Hospital
-
Using JMP® Pro 13 To Predict Depression Among Adolescents (US 2018 306)
Grant Reeves, Student, Oklahoma State University
Moumita Sen, Student, Oklahoma State University
Brandon Millikin, Student, Oklahoma State University
-
Using JMP® to Synthesize Computational Model Output and Test Data for an Army Weapon Physics of Failure Analysis (US 2018 316)
Douglas Ray, Lead Statistician, US Army ARDEC
Adam Foltz, Engineer, US Army ARDEC
-
Utilizing the Nonlinear Design Platform in JMP® to Design Chemical Reaction Kinetics Experiments (US 2018 315)
Kevin Singer, Statistician, US Army ARDEC
-
What’s Your Competitive Advantage? (US 2018 317)
Cyriac Wegman, President and Founder, SY64
