Transcript USCOTS 2015 Workshopx

USCOTS 2015
Exercises for teaching statistics with
simulations, resampling methods,
and big data
Mia Stephens, Julian Parris, Brant Deppa,
and Dick De Veaux
Copyright © 2010 SAS Institute Inc. All rights reserved.
What is JMP?
 Statistical Discovery Software from SAS
 Developed in 1989 as a desktop application
 Comprehensive
 basic statistics and graphical summaries
 advanced tools and techniques
 Visual, dynamic and interactive
 Extendible
 application and add-in builders
 integrates with Excel, R, MATLAB, and SAS
 Powerful programming language (JSL) for creating simulations,
custom applications and add-ins
 JMP 12, JMP 12 Pro, and JMP 12 Student Edition
2
Copyright © 2010, SAS Institute Inc. All rights reserved.
JMP 12 Student Edition
 Feature limited version of JMP, for high school and
introductory statistics
 Released mid-June
 Bundled with many textbooks, simple download process
 Many new features:
 No limit on number of rows
 Interactive Teaching Modules and Statistics Calculators
 Partition (Classification and Regression Trees)
 Variability (multi-vari) charts
 Save as HTML and Powerpoint
 Bootstrapping
3
Copyright © 2010, SAS Institute Inc. All rights reserved.
Workshop Schedule
Agenda (1:00 – 4:00):
1:00 Mia
Kickoff and Welcome
1:10 Julian This is JMP 12 (and JMP 12 SE)
1:30 Mia
Simulations and Resampling in JMP
2:00 Break
2:10 Brant
Exercises for Teaching Statistics with “Big” Data
2:55 Dick
Stat 101 Toolkit for Presenting Statistical Concepts
3:40 Group Q&A, and Panel Discussion
Materials are posted on the JMP Academic Community at:
https://community.jmp.com/docs/DOC-7471.
4
Copyright © 2010, SAS Institute Inc. All rights reserved.
Simulations and Resampling in JMP
 Built-in simulation and animation (p-value and power
animators, profiler with Monte Carlo)
 Legacy simulators (sample data and sample scripts)
 Custom simulations (formula editor, scripting)
 One-click bootstrapping (JMP Pro, and JMP 12 SE)
 Interactive teaching modules
 BU Statistic and Probability Lab
 Permutation Test for Two Means or Medians
 Randomization Test Add-in
5
Copyright © 2010, SAS Institute Inc. All rights reserved.
Built-In Simulations and Animation
Available since early releases of JMP:
 p-value animator and power animator
 Teaching Demonstrations and Simulators in the Sample
Scripts directory (confidence intervals, least squares,
correlation,…)
 Openly accessible code
 Modify and create your own
 Custom simulations using the random data generator
(Formula Editor)
 Prediction Profiler with Monte Carlo simulations
6
Copyright © 2010, SAS Institute Inc. All rights reserved.
P-Value Animator
7
Copyright © 2010, SAS Institute Inc. All rights reserved.
Custom Simulations
 Add rows to a data table
 Use Formula Editor to create function
(i.e. Random t(10))
8
Copyright © 2010, SAS Institute Inc. All rights reserved.
Custom Simulations
 Graph in Graph Builder (or Distribution) and explore
9
Copyright © 2010, SAS Institute Inc. All rights reserved.
Bootstrapping
10
Copyright © 2010, SAS Institute Inc. All rights reserved.
Bootstrapping
11
Copyright © 2010, SAS Institute Inc. All rights reserved.
Prediction Profiler
12
Copyright © 2010, SAS Institute Inc. All rights reserved.
Interactive Teaching Modules
 Bundled as free add-in (from JMP Academic Community)
 Built into JMP and JMP 12 SE
13
Copyright © 2010, SAS Institute Inc. All rights reserved.
Interactive Teaching Modules
14
Copyright © 2010, SAS Institute Inc. All rights reserved.
Permutation Test for Two Means or Medians
 New in sample data directory
 Based on code developed by Laura Schultz (Rowan)
 Launches a dialog - specify the columns of interest and
the number of resamples
15
Copyright © 2010, SAS Institute Inc. All rights reserved.
Permutation Test for Two Means or Medians
The script shuffles the levels of the factor and computes
averages after each reshuffle
Produces JMP table with original and resample averages,
and the difference.
• Saved scripts
produce graphical
results.
• Original sample is at
the bottom. Sort the
difference column to
find the empirical pvalue.
16
Copyright © 2010, SAS Institute Inc. All rights reserved.
Permutation Test for Two Means or Medians
Saved scripts allow you to produce graphical output of
results
17
Copyright © 2010, SAS Institute Inc. All rights reserved.
Randomization Test Add-In
Randomization Test Add-In (under development)
Current test options:
 Single quantitative outcome
 Two unpaired groups, quantitative response
 Single binary outcome
Randomization methods available:
 Resampling with replacement
 Shuffling group membership
 Shifting to the null with replacement
18
Copyright © 2010, SAS Institute Inc. All rights reserved.
Randomization Test Add-In
19
Copyright © 2010, SAS Institute Inc. All rights reserved.
Questions
How does using real/modern data change how you teach?
 Role of data preparation?
 Challenges of using real data?
 Finding real data?
What is the first thing you do when you open a data set in
the classroom?
 Where do you start? What do you do with the data?
Technology in the classroom
 What is the role of technology?
 What are the challenges of using technology?
 How has technology changed/shaped how you teach?
20
Copyright © 2010, SAS Institute Inc. All rights reserved.
Questions
What drives how/what you teach?
 Software?
 Technology?
 Industry?
How have the expectations you have of your students
changed (over, say, the past 10 years)?
 Introductory level
 More advanced
What is the role of simulations and resampling for
teaching statistical concepts?
21
Copyright © 2010, SAS Institute Inc. All rights reserved.
Jmp.com/tools
Jmp.com/teach
[email protected]
[email protected]
Thank you!
Copyright © 2010 SAS Institute Inc. All rights reserved.