|
Data defines the model by dint of genetic programming, producing the best decile table.
|
|
Webcast for Demonstrating the GenIQ Model
Bruce Ratner, Ph.D. |
|
|
Subject: WEBCAST for Statistical Modelers and Data Miners: Demonstration of the GenIQ Model© - no charge!
Date: Mutually Convenient Date & Time
The GenIQ Model© is a machine learning alternative model to the statistical ordinary least squares and logistic regression models. GenIQ lets the data define the model – automatically data mines for new variables, performs variable selection, and then specifies the model equation – so as to "optimize the decile table," to fill the upper deciles with as much profit/many responses as possible. Put differently, GenIQ seeks to maximize cum lift, a measure of model predictiveness of identifying the upper performing individuals often displayed in a decile table. GenIQ produces models that outdo statistical models, and is a different model: unsuspected equation, ungainly interpretation, and easy implementation. GenIQ requires no programming, produces models that outdo statistical models, and is a different model: unsuspected equation, ungainly interpretation, and easy implementation.
Prerequisites for the GenIQ Webcast
1. Entry-level/well-schooled modeler of statistical ordinary least squares and logistic regression models, AND who can think out of the box.
OR
2. Mid-level or Advanced modeler of statistical ordinary least squares and logistic regression models, AND who can think out of the box.
GenIQ Webcasts are one-on-one, you, your invited colleagues, and me.This allows for me to give you my full attention to answer all your questions, without feeling that you are taking-up other attendees' time. Please do not be put off by the webcast. It has proven to be the most effective way of my explaining the GenIQ Model.
After the webcast, I email you the GenIQ demo download link (*) and the GenIQ-webcast pdf-file for your reference.
To Register one of the webcast versions (described below), please anwser the multipart Question, written lower on this page:
Version 1: The GenIQ Model Software Demonstration (30 minutes; must have an understanding of genetic programming (GP)).
Version 2: The GenIQ Model Presentation (90 minutes: GP theory-lite tutorial and demonstration of the GenIQ software).
Verision3: If a 90-minute webcast is not practical due to your workload, then I can split the webcast into two sessions:
The first webcast is be a 60-minute GP theory-lite tutorial, and the second one is be a 30-minute demonstration of the GenIQ software.
The GenIQ Model Webcast is a perfect place to start if you want to perform better predictive modeling and data mining!
Question: First, please look over the items (below), and then answer them by clicking here.
- Your statistical modeling experience: a) entry-level, b) mid-level, or c) advanced modeler.
- Your typical model has predictor variables: a) less than 6, b) 6 to 12, c) 13 to 20, or d) greater than 20 predictor variables.
- You transform predictor variables (e.g., 1/x, log x): a) always, b) sometimes, or c) never.
- Your method of variable selection: a) forward selection, b) backward elimination, c) stepwise, or d) other, please specify.
- Your GP experience: a) none, b) have a basic sense of GP terms: i) population size {models/generation}, ii) operators {copying, mating, mutation}, or c) have taken a GP course.
- Do you, at least in part, use the decile table/analysis to determine the best model? a) yes, or b) no.
- Two dates & times available for the GenIQ Webcast.
- You would like to attend: a) The GenIQ Model Software Demonstration, b) The GenIQ Model Presentation, or c) Two-session GenIQ Webcast.
- Thinking out of the box on a 10-point scale: 1 = you never consider new ideas, ... , 10 = you are always willing to consider new ideas.
- GenIQ Testing Agreement: If I request (to ensure GenIQ is properly validated; of couse I would sign a NDA), would you provide the data used in building and comparing GenIQ and your model? a) yes, or b) no.
* GenIQ Testing: a) within the first week after webcast, b) after current project is done; guestimate mm/dd, or c) not sure when.
Once we agree on a date & time, you will receive a "GoToMeeting© Invitation" email with instructions on how to join the event. I discuss:
- Data basics: What kind of data is required for genetic modeling and data mining; in what format must the data be; what preprocessing steps are necessary (There are none! And, I explain why).
- What kinds of questions can be answered with GenIQ data mining.
- How GenIQ models work: The inputs, the outputs, and the nature of its predictive mechanism of GP.
- Evaluation criteria: How GenIQ predictive models can be assessed and their worthiness measured.
I guarantee:
- To show that GenIQ lets the data define the model – automatically data mines for new variables, performs variable selection, and then specifies the model equation – so as to optimize the decile table/cum lift.
Please do not hesitate to contact me if you have any questions.
Bruce Ratner, Ph.D.
|
| For more information about this article, call Bruce Ratner at 516.791.3544 or 1 800 DM STAT-1; or e-mail at br@dmstat1.com. |
|
|
br@dmstat1.com