2018 JMP School on Statistics

An important step in any scientific study is to summarize and visualize the available data. This course shows that the JMP software is the ideal tool to organize and re-organize data, to create summary tables and generate a multitude of graphical representations in the blink of an eye. At least as important is the fact that JMP is highly interactive, so that it allows exploration and pattern detection interactively and painlessly. Throughout this hands-on course, the participants will have ample opportunity to familiarize themselves with JMP and its capabilities for summarizing and visualizing data. Part of this course will be based on the book “Statistics with JMP: Graphs, Descriptive Statistics and Probability” by Peter Goos and David Meintrup.

Analysis of variance (ANOVA) and linear regression are the two most fundamental statistical models used in numerous applications. Every scientist, engineer or economist who works with data will most likely encounter these modeling techniques in his area of expertise. This course will provide everything that is needed to understand and apply ANOVA and linear regression models. We will start with introducing three key concepts of statistical inference: point estimators, confidence intervals, and hypothesis tests. Next we will study a wide range of examples of hypothesis tests, reaching from tests for one and two populations to nonparametric tests. For more than two populations, we will introduce the analysis of variance, investigate its use and limitations. Finally, we will turn our attention to linear regression models and how to apply them to given data sets. All parts of this course will be accompanied by examples, case studies, and exercises that will be performed with the highly interactive and user-friendly software JMP. The content of the course will be based on the book “Statistics with JMP: Hypothesis Tests, ANOVA and Regression” by Peter Goos and David Meintrup.

Often, we want to perform a regression analysis in situation where the response is not a normally distributed continuous variable. Sometimes, the response is not even quantitative. This course focuses on a family of regression models that allows for binary responses, ordinal categorical responses, responses that are counts and continuous but non-normally distributed responses. The best known of these models are logistic regression model and Poisson count models. Both of these models belong to the family of generalized linear models, which generalizes the standard linear regression model. This course first provides an introduction to generalized linear models in general. Next, it discusses several useful models within that family. We pay special attention to the graphical tools provided by the JMP software to visualize the impact of the explanatory variables and use examples from different disciplines to demonstrate the broad applicability.

Modern data sets often involve many variables and turning such data sets into information generally requires multivariate statistical analyses. In this course, the most commonly used multivariate statistical analyses are discussed. First, we will demonstrate methods that allow for a quick visualization of multivariate datasets, and to identify outlying observations.  Once the clean dataset is obtained, several methods will be discussed to analyze these data.  Techniques discussed will cover most common practical situations, such as making predictions, finding groupings in the data, and classifying observations.  Basic principles of these methods will be explained, after which they will be applied during a JMP hands-on session.