Statistical Modeling Week 2011
|October 17 - 21, 2011, Boston
|Bridging the Gap Between Theory and Practice - Our 34th Year!|
|Statistical Modeling Week is an annual conference sponsored by Statistical Innovations Inc., featuring applications-oriented seminars focusing on the latest trends in statistical analysis. Led by a group of experts in the field, the seminars teach professionals how to apply and understand today's major breakthroughs in statistical methodology.|
Statistical Modeling Week 2011 will take place in Boston on October 17 - 21, 2011.
Whether you work in marketing, advertising, health, econometrics or evaluation, you will learn to apply these important breakthroughs to your everyday work. And you'll learn more than the techniques - you'll learn the theories behind them.
Numerous leading organizations have taken advantage of our prior seminars -
here is a sample list.
Attendance is limited. Register for Statistical Modeling Week 2011 today!
Course descriptions for 2011 will be available shortly. Below are the course descriptions for Statistical Modeling Week 2010.
You can also download our 2010 conference brochure.
| Tuition and Enrollment Information|
October 18, 2010. Predictive Modeling with Many Correlated Predictors ||$895|
October 19 - 20, 2010. Latent Class and Finite Mixture Modeling ||$1,490|
October 21 - 22, 2010. Modeling Consumer Decision Making and Discrete Choice Behavior ||$1,490|
Note: Discounts are available for multiple course registrations. After enrolling in the first course for an undiscounted tuition, you deduct $50 for each additional course you sign up for (the maximum discount would be $100 if all 3 courses are taken).
Discounts are also available for additional people who sign up from your organization. Their tuition is $795 for each 1-day seminar and $1,390 for each 2-day seminar.
Please note that the discounts will not appear on your order form when you register. We will include the discounts when we send you the invoice for the sessions.
| Predictive Modeling with Many Correlated Predictors |
Who Should Attend
Marketing, biomedical and other researchers who want to improve their understanding of regression model development in the presence of many correlated predictors.
Familiarity with linear and logistic regression analysis at an applied level.
Recent advances in analysis of high dimensional data now allow reliable regression models to be developed even when the number of predictors exceeds the number of cases! In this course we begin by reviewing problems and limitations with traditional linear and logistic regression. Our applications-oriented presentation provides insight into how the new approaches work through examples and by providing an overview of the relevant theory, supplemented by the supporting equations. We use real and simulated data sets to illustrate the different approaches.
- Traditional Predictive Modeling
- Continuous vs. Categorical Dependent Variables
- Linear Regression
- Logistic Regression and ROC Curves
- Discriminant Analysis
- Model Selection Criteria
- p-values and Degrees of Freedom
- Information criteria: AIC, BIC
- Correlated Predictors and Multicolinearity
- 10-fold cross-validation
- Variable Reduction
- New Methods and Software for Dealing with Multicolinearity and Overfitting
- Biased and Unbiased Regularization and Model Averaging Methods
- Lasso and Ridge Regression
- Principle Component and PLS Regression
- Variable Reduction – Least Angle Regression (LARS)
- Simultaneous Model Estimation and Variable Reduction
Examples with Correlated Component Regression (CCR) and CORExpress™
- Enhancing Model Performance with Suppressor Variables
- Scree Plots and Coefficient Path Plots
- Generalized Degrees of Freedom (GDF) and Related Statistics
What you will learn
Back to Courses
- How to develop reliable models, even when with extreme multicolinearity and when # predictors > n
- Why many popular variable selection techniques are suboptimal
- About a new powerful step-down variable reduction technique in CORExpressTM
- How GDFs shrink inflated statistics to adjust for overfitting
- About free and commercially available software for handing high dimensional data
| Latent Class and Finite Mixture Modeling |
Interest in Latent Class (LC) models is increasing rapidly. This is due to the growing evidence that they provide better solutions than traditional approaches to cluster, factor and regression analysis when the population is not homogenous. In this 2-day course we introduce LC as a probability model and describe various applications using the latest version of Latent GOLD®. On day 1, we focus on model fitting strategies and the interpretation of output. On day 2, we consider several advanced topics including identification issues, random effects with continuous factors, repeated observations, hidden (latent) Markov models for latent growth, and the computation of individual-level coefficients.
- Basic ideas of latent class analysis
- The concept of local independence
- The general probability model
- Handling nominal, ordinal, continuous and count variables
- LC measurement models
- Discrete vs. continuous factor analysis
- LC regression and segmentation models
- Inclusion of covariates in LC models
- Model-based clustering / latent discriminant analysis
- Comparing models and assessing fit
- Uni-plot, bi-plot, tri-plot and profile plot
- Case studies and computer demos
- Model specification using LG-EquationsTM
- Identification problems and boundary solutions
- Use of Bayes constants to eliminate boundary solutions
- The problem of local solutions
- Use of randomized start values
- Use of the bootstrap p-value
- Bivariate residuals to diagnose local dependencies
- Repeated measures / conjoint marketing
- LC Growth models and latent transition models
- Use of continuous factors as a random intercept
What you will learn
Back to Courses
- How to specify LC cluster, factor and regression/segmentation probability models.
- Interpreting graphical displays of results.
- What to look for when interpreting output.
- Strategies for assessing model fit with sparse and non-sparse data.
- How to isolate the scale effects in ratings data.
- How to include nominal, ordinal and continuous variables in the models.
- Why LC models improve over K-means clustering.
| Modeling Consumer Decision Making and Discrete Choice Behavior |
Who Should Attend
This course is intended for researchers and advanced practitioners in marketing, economics, and fields in which consumer demand and choice is of interest.
Participants should have a background in statistics and some familiarity with econometrics, but advanced training is not necessary.
The workshop covers a wide array of related topics in the design of discrete choice experiments (DCEs), the analysis of data collected from these experiments and recent extensions of the theory of Best-Worst Scaling (BWS). The emphasis is on breadth, with sufficient depth provided to allow participants to learn the basics and learn how to further their understanding. Hands-on examples are provided and used as a basis for teaching basics, allowing participants to understand not just conceptual ideas, but also the mechanics of implementing the ideas in practical settings of interest to academics and practitioners. The workshop deals with design and implementation of DCEs and BWS surveys. It covers design approaches, statistical models for analyzing data and developing predictive models, limitations and unresolved issues. It discusses good and bad practices, myths and evidence. The emphasis is on the science and associated empirical evidence, not opinion, anecdotes, “experiences” and the like.
- Basics of experimental design & example DCEs
- Designing DCEs – experimental design concepts
- Types of discrete choice experiments by application
- Participants design and analyze a DCE
- More advanced DCE topics 1 – individual difference
- More advanced DCE topics 2 – modeling individuals
- More advanced DCE topics 3 – data pooling & validity
- Participants design a DCE for single persons
- Introduction to BWS – overview & Case 1
- Introduction to BWS – Case 2
- Introduction to BWS – Case 3
- Participants design and analyze a BWS survey
What you will learn
Back to Courses
- About discrete choice experiments (DCEs) and associated models for analyzing data collected in these experiments
- How to design different types of DCEs.
- The difference between “top-down” and “bottom-up” ways of analyzing choice data and developing models, including models for single persons.
- About new developments in Best-Worst Scaling (also called “Max-Diff Scaling”), including cases 1, 2 and 3 that allow one to rigorous measure and model a wide array of things typically modeling with rating scales, paired comparisons and DCEs; learn about the critical role that non-constant error variances play in estimation and interpretation of choice models.