SOCR: Statistics Online Computational Resource

Ivo D. Dinov, PhDProfessor and SOCR Director, Dinov@ucla.edu

www.StatisticsResource.org

It’s Online, Therefore It Exists!

1. SOCR Overview The Statistics Online Computational Resource

1. SOCR Overview The Statistics Online Computational Resource

•

Challenge: Provide integrated, cost-effective,

pedagogically-relevant, IT-enhanced, multidisciplinary,

multilingual & data-driven science educational resources• Web-based interactive applets• Multilingual learning activities• Instructional resources• Training

(students/teachers)• Data-driven

Science Education

•

SOCR Approach:• Open web development & utilization• Integrated research, data, tools & learning

•

Why?• Reduce the Digital Divide

(age, geographic, socio-economic)• Improve learning experiences and merge research & learning

InstructionalResources

ToolsApplets

InstructorWorkshops

Data

LearningActivities

StudentTraining

2.SOCR User Types & Their Needs2.SOCR User Types & Their Needs

• Audiences•Middle-High school + College•Formal learners (courses)• Informal learners (refreshers, tutorials, EBooks)•Science learners (starting with probability,

statistics and math education)

• Current Barriers •Non-interoperable/disconnected learning resources•Within-discipline & language-bound curricula•Cost: print books, science tools, pay-per-view sites

• Characteristics of SOCR Users•User Volume: 1 Million page views (last 12 mo’s)•Types of User Resources: EBooks and Applets•User Growth: >20% annually (since 2005)•Advanced Placement (Stats) courses (10K in US): 30%•College Classes (100-5K in US):30%• Informal users (incl. industry): 40%

ForumNavigator

3. SOCR Solutions & Resources3. SOCR Solutions & Resources

• Web-Applets•Open-source development (LGPL license): >500 apps•http://SOCR.googlecode.com/•Wide spectrum of cross-disciplinary applets•Educational use with Research applications

• Interactive Learning•Hands-on activities (Wiki-based): > 1,200 Wiki pages•Auto translated in 24 different languages•

http://wiki.stat.ucla.edu/socr/index.php/SOCR_EduMaterials• Integrated concepts, data, methods, & technology•Research-derived, simulated & observational Data

• Instructional Resources•EBooks, e.g., http://wiki.stat.ucla.edu/socr/index.php/EBook•Training (mentoring students; teacher continuing Ed)•Resource Navigators:

http://socr.ucla.edu/SOCR_HT_ResourceViewer.html

ToolsApplets

Workshops

Data

LearningActivities

InstructionalResources

Training

4.Alternative Solutions & Providers (Barriers to Adoption)

4.Alternative Solutions & Providers (Barriers to Adoption)

• Big Dogs •

Science Book Publishers (print/pay-per-view)•

Wikipedia/Wikibooks

(content only, no tools/training)•

Minitab (>$100/yr), Mathematica

($150/yr)•

StatCrunch

($20/yr), JMP ($40/yr), Calculators•

Statistical Computing Package R

(free, tools only)• Inertia

•

Institutional (rigid curricula/slow changes)•

Teachers/Instructors (lack of IT expertise/time)•

Publishers (financial interests, support status quo)• Innovators

•

For-Profits (e.g., WolframAlpha)•

Non-profits (future NSF/NIH funded initiatives)• Surpass Barriers

•

SOCR promotes resource interoperability

& customization

(subject, syllabus, audience, level)•

Learners are better then (most) instructors at finding/using modern IT-enhanced resources

SOCR Approach Benefits•

SOCR developments are translational

– driven

by learner needs, implemented by students, tested & validated in the classroom

•

SOCR resources (code, materials, tools, activities) are community peer-reviewed

•

SOCR Core Principle: If it’s not online, it does not exist!

•

SOCR adopts the CreativeCommons/Wikipedia open-knowledge

model•

SOCR is constantly reviewing

& adopting

new Information & Communication Technologies to improve user experiences

•

Feedback

from trainees, students & the general community is overwhelmingly exiting & supportive

•

SOCR alternatives

require paid subscriptions to multiple sites, print materials, computational resources, special installations

•

SOCR delivery mechanism is scalable

& distributed

(all calculations carried on user machines –

only instructions are transferred)•

Free, anonymous, multilingual access –

no barriers

5.

SOCR Resource Development & Dissemination

5.

SOCR Resource Development & Dissemination

•Development cycle•

Identify Needs

& draft Design Specs•

Tool

Implementation, Continuing Testing & Docs•

Activity

generation and resources integration•

Classroom validation

and resource peer-review•

Wide web-based dissemination

& Instructor training• Dissemination

•

Learner/Instructor awareness of new SOCR resources•

Peer-review publications, National/International

Conference presentations, Organizing Training

Workshops, Webinars

•

User-guides, EBooks, Forum, anonymous surveys•

Multiple SOCR servers & Mirror servers•

No software installation necessary

–

only a Java enabled web-browser is required

•

Easy anonymous access/testing

of all SOCR Resources via the Internet or via local (JAR/HTML) download

•

Resource Navigators http://socr.ucla.edu/

SOCR Servers

•

Main SOCR ServersMain http://socr.ucla.eduWiki http://wiki.stat.ucla.edu/socrForum http://forums.stat.ucla.edu/socrGoogleCode

http://socr.googlecode.com

•

Mirror ServersAm Stat Assoc

http://www.amstat.org/publications/jse/socrPsyresearch

http://psyresearch.org/statistics/socrNational Science Digital Library (NSDL)http://nsdl.org/collection/mathematics/

Example of End-to-End Science Ed using SOCR Resources:

A. Problem Identification

Example of End-to-End Science Ed using SOCR Resources:

A. Problem Identification

CA Ozone Pollution

•

Are there temporal changes in California Ozone?

•

What is the geographic

distribution of the California Ozone pollution and is it changing with time?

Example of End-to-End Science Ed using SOCR Resources:

B. Gather Data

Example of End-to-End Science Ed using SOCR Resources:

B. Gather Data

CA Ozone Pollution

http://wiki.stat.ucla.edu/socr/index.php/SOCR_Data_121608_OzoneData

INDEX VARIABLE LOCATION YEAR MTH_1 … MTH_12 ANNUAL

1 OZMAX1HR 2008 1980 0.07 … 0.09 0.12

2 OZMAX1HR 2008 1981 0.07 … 0.1 0.11

3 OZMAX1HR 2008 1982 0.07 … 0.09 0.15

4 OZMAX1HR 2008 1983 0.06 … 0.07 0.14

5 OZMAX1HR 2008 1984 0.07 … 0.07 0.14

6 OZMAX1HR 2008 1985 0.09 … 0.09 0.13

7 OZMAX1HR 2008 1986 0.1 … 0.07 0.11

… … … … … … … …

Example of End-to-End Science Ed using SOCR Resources:

C. Background Research

Example of End-to-End Science Ed using SOCR Resources:

C. Background ResearchCA Ozone Pollution

Example of End-to-End Science Ed using SOCR Resources: D. Data Interrogation

Example of End-to-End Science Ed using SOCR Resources: D. Data Interrogation

CA Ozone PollutionSOCR MotionChartshttp://socr.ucla.edu/SOCR_MotionCharts/

Example of End-to-End Science Ed using SOCR Resources:

E. Data Analysis

Example of End-to-End Science Ed using SOCR Resources:

E. Data AnalysisCA Ozone PollutionSOCR Analyseshttp://socr.ucla.edu/htmls/SOCR_Analyses.html

1980 2006

Example of End-to-End Science Ed using SOCR Resources:

F. Inference & Decision-Making

Example of End-to-End Science Ed using SOCR Resources:

F. Inference & Decision-Making

What is the geographic distribution of the California Ozone pollution and is it changing with time?

1980 2006

> 0Are there temporal changes in California Ozone?

OzonePollution

= -191.6095127656 + 0.09667 ×

YEAR

6.

Sustainability of the SOCR Open Collaborative Model

6.

Sustainability of the SOCR Open Collaborative Model

Goals (next 3-5 yrs)• Increase proactive dissemination efforts (learners,

instructors & general industry users)• Extend Prob, Stats, Math framework to Eng, Bio,

Neurosci, Phy, Chem

(Multidisciplinary Science Ed)• Increase computational library• Improve and Extend the Learning Activities• Engage outside instructors in resource validation &

classroom integration (OSU, NCSU, TAMU, etc.)

Sustainability• Is the CC/Wikipedia model sustainable

(short/midterm)?• Potential development risks (lack of funding for

development & active dissemination)• Potential for scooping by for-profits (SOCR is

completely open –

dev, know-how & usage)

SOCR Partnerships

•

UCOP –

development of Advanced Placement K-12 curricular materials

•

Consortium for the Advancement of Undergrad Stats Education (CAUSE) –

sharing existing & developing new learning resources

•

International Statistical Literacy Project (ISLP)

•

Multimedia Educational Resource for Learning and Online Teaching (MERLOT)

•

SOCR resources used in 8 (outside) books

•

Many (anonymous) courses across the Globe utilize SOCR (indirect ISP monitoring)

http://wiki.stat.ucla.edu/socr/index.php/SOCR_Partners

7.

Financial Projections & Outcomes

7.

Financial Projections & Outcomes

Outcomes of SOCR Resources•

Improved student motivation & participation•

Enhanced learning experience by using modern interactive tools and technologies

•

Improved scientific skills of workforce•

What investment is required to carry the company to the next major level of valuation?

•

Reduced Digital/Language/Access Divide•

UCLA Statistics will maintain perpetually all SOCR resources (but does not have funds to upgrade servers or support development or training efforts)

•

SOCR is a non-for-profit resource•

The general societal benefits of SOCR developments are realized as improvements of open-access multidisciplinary science education for all

2009-10 2010-11 2011-12 2012-13 2013-14New Users 300K (exp)400K (exp)500K (exp)650K (exp) 1 MNew Tools 15 15 20 20 20Improved Tools 12 15 15 20 20Learning Materials 20 25 25 30 30Revenues $120K $80K (exp)$80K (exp)$80K (exp)$80KExpenditures $125K $125K $135K $130K $135K

Development $70K $70K $70K $70K $70KStudent Stipends $40K $40K $40K $40K $40K

Travel/Dissemination $10K $15K $20K $20K $20KHardware Upgrade $5K $5K $5K

Projections

8. SOCR Funding Support &

Advisory Needs

8. SOCR Funding Support &

Advisory Needs

SOCR Personnel Expenses

•

SOCR faculty

(3 UCLA Faculty): already committed > 10,000’s hours to the project (at no cost), will continue to contribute as part of their faculty research, teaching & service activities

•

SOCR students

(rotating 2 grads & 8 undergrads) –

2-3 students are partially funded, the rest are vols

•

SOCR programmer analyst

(PA) –

software developer & server administrator (100%) on NSF grant

•

35 Educators

selected (US) & invited to attend annual 3-day SOCR Continuing Ed workshops (free training & community dissemination)

2010-11 2011-12 2012-13 2013-14Revenues $80K (exp)$80K(exp)$80K(exp)$80K

NSF Grants $80K (exp)$80K (exp)$80K (exp)$80K

Expenditures $125K $135K $130K $135KFaculty $0K $0K $0K $0KPA Dev $70K $70K $70K $70K

Student Stipends $40K $40K $40K $40KTravel/Dissemination $15K $20K $20K $20K

Hardware Upgrade $5K $5K

Deficit (Projected) $45K $55K $50K $55K

SOCR project also needs advisory, development & assessment support

from information &communication technology experts

Value Added & Derived Benefits

Value Added & Derived Benefits

Value Added•

SOCR infrastructure provides end-to-end solutions

for developing and disseminating Science Ed tools, learning materials & instructional resources

problem data model analysis inference•

Only a Java-enabled browser

is required for browsing, traversing, navigating, utilizing,

customizing or integrating all SOCR resources

•

In Science Ed, one-size-doesn’t-fit-all

–

SOCR provides the core building blocks for learners & educators (resource customization, search & navigation)

•

Long-term sustainability dependent on community involvement, institutional commitments &

continuous integration of modern IT resources

Benefits Delivered•

SOCR reduces the age, language, geographic, & socio-economic Barriers to multidisciplinary science education

•

SOCR resources significantly improve learning experiences & knowledge retention for all formal and informal learners

•

SOCR facilitates dynamic science curricular development & IT-blended instruction•

SOCR resources aid the on-the-job training

& knowledge refreshing

for the industry workforce

Supplementary MaterialsSupplementary Materials

Examples of SOCR Use: A. Science Education

Examples of SOCR Use: A. Science Education

SOCR Resources are rigorously tested

in (randomized) design-experiments• Lower/Upper/Graduate CoursesMajor, Minor & Service CoursesSmall (<10) & Large Classes (160)20 Instructors> 20 courses> 2,000 students

•Findings•Statistically significant SOCR-treatment effect

(measured by quantitative exams) on student learning•Strong consistency of SOCR-treatment benefit (across

tests, courses, topic, levels)http://www.socr.ucla.edu/htmls/SOCR_References.html

Demographics Stat13.1 Control (Traditional)

Stat13.2 SOCR-Treatment

Freshmen 24 7Sophomores 18 14

Juniors 16 38Seniors 23 29

Graduates 2 0Total 83 88

Index of Exam Questions

Qua

ntita

tive

perfo

rman

ce

SOCR-Treatment Class

Control Traditional Class

Examples of SOCR Use: B. Translational Biomedical Science

Examples of SOCR Use: B. Translational Biomedical Science

SOCR Resources are employed in computational neuroscience•

Largest to-date study of association between cognitive performance, brain structure, dementia

& an obesity gene

(FTO), PNAS’2010Relations between obesity gene (FTO) and brain volume reduction in the elderly. Subjects with higher body-mass index (BMI) had significantly lower regional brain volumes in many areas. With every 1-unit increase in BMI, there was an associated 1–1.5% average brain tissue reduction

in frontal, temporal, parietal, and occipital lobe regions.

•Brain mapping

of prenatal exposure

to Methamphetamine and Alcohol using Tensor-

Based Brain Morphometry

& SOCR AnalysesStatistical maps differentiating locally the brains of children with prenatal exposure to methamphetamine and alcohol versus alcohol and no-

methamphetamine use. (1) CON < MAA (red), (2) CON < MAA and CON < ALC (orange), (3) CON < MAA and ALC < MAA (yellow), (4) CON < ALC, CON < MAA, and ALC < MAA (green), (5) CON > MAA and ALC > MAA (dark blue), (6) CON > MAA (light blue), (7) CON > ALC and CON > MAA (purple), and (8) CON > ALC, CON > MAA, and ALC < MAA (light pink).

http://www.socr.ucla.edu/htmls/SOCR_References.html

Core SOCR ResourcesCore SOCR Resources

Tools & Activitieshttp://www.SOCR.ucla.edu

wiki.stat.ucla.edu/socr/index.php/SOCR_EduMaterials

Virtual Demos http://www.SOCR.ucla.edu

Concepts& Methods

http://wiki.stat.ucla.edu/socr/index.php/EBook

Datahttp://wiki.stat.ucla.edu/socr/index.php/SOCR_Data

Infrastructurehttp://socr.ucla.edu/SOCR_HT_ResourceViewer.htmlhttp://wiki.stat.ucla.edu/socr/index.php/SOCR_News

http://socr.ucla.edu/htmls/SOCR_Languages.htmlhttp://socr.ucla.edu/docs/SOCR_Documentation.html

SOCR DistributionsSOCR DistributionsBernoulli Distribution; Beta Distribution; Beta (Generalized) Distribution; Binomial Distribution; Birthday Distribution; Cauchy Distribution; Chi-

Square Distribution; Circle Distribution; Continuous Uniform Distribution; Die Distribution; Discrete ArcSine

Distribution; Discrete Uniform Distribution; Erlang

Distribution; Error

Distribution; Exponential Distribution; Fisher's F Distribution; Fisher-Tippett

Distribution; Gamma Distribution; General Cauchy Distribution;

Geometric Distribution; Gilbrats

Distribution; Gumbel

Distribution; Half-Normal Distribution;

HyperGeometric

Distribution; Laplace Distribution; Logarithmic Distribution; Logistic

Distribution; Log-Normal Distribution; Matching Distribution; Maxwell Distribution; MixtureDistribution; Negative-Binomial Distribution; Normal Distribution; Pareto

Distribution; Point-Mass Distribution; Poisson Distribution; Poker-Dice Distribution; Power-

Function Distribution; Rayleigh Distribution; Student's T Distribution; Student's T Non-Central Distribution; Triangle Distribution; Von Mises

Distribution; WalkMaxDistribution; WalkPositionDistribution; Weibull

Distribution;

FEATURES

•Largest Collection

•75+ Distributions

•Graphs

•Density functions

•Cumm

Distr

Func’s

•Inverse CDFs

•Random Sampling

•Moments

•Web Interface

http://www.socr.ucla.edu/htmls/dist/

SOCR ExperimentsSOCR Experiments

FEATURES

•70+ Experiments

•Simulations

•Summary Stats

•Model vs. Observ.

•GUIs

•Web Interface

•Based on VLPS

http://www.socr.ucla.edu/htmls/exp/

Ballot Experiment Ball and Urn Experiment Bertrand Experiment Beta Coin Experiment Beta Estimate Experiment Binomial Coin Experiment Binomial Timeline Experiment Birthday Experiment Bivariate

Normal Experiment Bivariate

Uniform Experiment Buffon's Coin

Experiment Buffon's Needle Experiment CardExperiment

Chi Square Dice Experiment Chuck A Luck Experiment Coin Die Experiment Coin Sample Experiment Confidence

Interval Experiment Coupon Collector Experiment Craps Experiment Dice Experiment Dice Sample Experiment Die Coin Experiment Finite Order Statistic Experiment Fire

Experiment Galton Board Experiment Game Gamma Estimate Experiment Gamma Experiment Markov Chain

Experiment Match Experiment Mean Estimate Experiment Mean Test Experiment Mixture Model EM Experiment Monty Hall Experiment Negative Binomial Experiment Normal

Estimate Experiment Order Statistics Experiment Pareto Estimate Experiment Problem of Points Experiment Two-

Dimensional Poisson Experiment Poisson Experiment Two-

Type Poisson Experiment Poker Dice Experiment Poker Experiment Probability Plot Experiment Proportion Estimate Experiment Proportion Test Experiment Quantile

JApplet

Random Variable Experiment Randowm

Walk Experiment Red and Black Experiment Roulette

Experiment Sample Mean Experiment Sign Test Experiment Spinner Experiment Triangle Experiment Uniform Estimate Experiment Variance Estimate Experiment Variance Test Experiment Voter Experiment

SOCR AnalysesSOCR Analyses

FEATURES

•Param+NonParam

•Graphs

•Summary Stats

•R Interface

•GUIs

•Web Interface

FEATURES

•Param+NonParam

•Graphs

•Summary Stats

•R Interface

•GUIs

•Web Interface

http://www.socr.ucla.edu/htmls/ana/

ANOVA -

One Way ANOVA -

Two Way 2

Model Goodness-of-Fit Test

Multiple Regression Analysis One Sample T Test

Simple Regression Analysis Two Independent Sample T Test

Two Independent Sample Wilcoxon

Rank Sum Test

Two Paired Sample Sign-Test Two Paired Sample Signed-Rank

Test (Wilcoxon) Two Paired Sample T Test

RESULTS:

Sample size=19

INDEPENDENT = GroupDEPENDENT = Dependent

DF Model = 2DF Error = 16DF Corrected Total = 18RSS MODEL = 45030.94956140351RSS ERROR = 2714.2083333333335RSS TOTAL = 47745.15789473685MSS MODEL = 22515.474780701756MSS ERROR = 169.63802083333334F-VALUE = 132.72658257916632P-VALUE = 1.0907141856364433E-10

SOCR ModelerSOCR ModelerFEATURES

•Largest collection of free models

•Distribution Model Fitting

•Fourier and Wavelet Data Modeling

•Random Number Generator (any SOCR distribution)

•Graphs

•GUIs

•Web Interface

FEATURES

•Largest collection of free models

•Distribution Model Fitting

•Fourier and Wavelet Data Modeling

•Random Number Generator (any SOCR distribution)

•Graphs

•GUIs

•Web Interface

http://www.socr.ucla.edu/htmls/mod/

BetaFit_ModelerExponentialFit_Modeler

FourierFit_ModelerGammaFit_ModelerMixedFit_ModelerNormalFit_ModelerPoissonFit_ModelerWaveletFit_Modeler

…

SOCR ChartsSOCR Charts

FEATURES

•60+ Dynamic Interactive Graphs

•Summary Stats

•GUIs

•Web Interface

•Based on JFreeCharts

FEATURES

•60+ Dynamic Interactive Graphs

•Summary Stats

•GUIs

•Web Interface

•Based on JFreeCharts

http://www.socr.ucla.edu/htmls/chart/

SOCR UsageSOCR UsageVisitor Log for Statistics Online Computational Resource

•> 2,600,000 active users•National and International Users•>20% Annual•

Summary