MOTIVATING INTRODUCTORY COMPUTING WITH PEDAGOGICAL DATASETS Austin - PowerPoint PPT Presentation

MOTIVATING INTRODUCTORY COMPUTING WITH PEDAGOGICAL DATASETS Austin Cory Bart Computer Science Applications, Virginia Tech March 22, 2017 1

Thanks! Clifford A. Shaffer Eli Tilevich Brett Jones Dennis Kafura Phill Conrad And many others! 2

Research Question “Can a Data Science context motivate introductory computing students, particularly non-Computing majors ?” 3

Contributions • A model for characterizing student motivation with respect to course components • New technology to support data science as an introductory computing context • A large collection of real-world datasets for non-computing majors • Evidence for value of a data science context as a motivating course component • Evidence that connects course content with engagement outcomes 4

Publications A. C. Bart, R. Whitcomb, E. Tilevich, C. A. Shaffer, D. Kafura, Computing with CORGIS: Diverse, Real-world Datasets for. 1. Introductory Computing (Best Paper) , SIGCSE '17, Seattle, Washington. March, 2017. D. Kafura, A. C. Bart, B. Chowdhury, Design and Preliminary Results From a Computational Thinking Course . ITiCSE'15, 2. Vilnius, Lithuania. July 6-8, 2015. A. C. Bart, J. Riddle, O. Saleem, B. Chowdhury, E. Tilevich, C. A. Shaffer, D. Kafura, Motivating Students with Big Data: 3. CORGIS and MUSIC , Splash-E '14, Portland, Oregon. October 21-23, 2014. A. C. Bart, E. Tilevich, T. Allevato, S. Hall, C. A. Shaffer, Transforming Introductory Computer Science Projects via Real- 4. Time Web Data , SIGCSE '14, Atlanta, Georgia. March 5-8, 2014. A. C. Bart, E. Tilevich, C. A. Shaffer,T. Allevato, S. Hall, Using Real-Time Web Data to Enrich Introductory Computer 5. Science Projects , Splash-E '13, Indianapolis, Indiana. October 26-31, 2013. (Related Publications) A. C. Bart, J. Tibau, E. Tilevich, C. A. Shaffer, D. Kafura, Design and Evaluation of Open-access, Data 1. Science Programming Environment for Learners , IEEE Computer '17. May, 2017 (accepted). A. C. Bart, J. Tibau, E. Tilevich, C. A. Shaffer, D. Kafura, Implementing an Open-access, Data Science Programming 2. Environment for Learners , COMPSAC '16, Atlanta, Georgia. June 10-15, 2016. A. C. Bart, C. A. Shaffer. Instructional Design is to Teaching as Software Engineering is to Programming . SIGCSE '16. 3. Kansas City, MO. March 2-5, 2016. A. C. Bart, E. Tilevich, C. A. Shaffer, D. Kafura, Position Paper: From Interest to Usefulness with BlockPy, a Block-based, 4. Educational Environment , Blocks & Beyond '15, Atlanta, Georgia. October 21-23, 2015. 5

Overview Motivation Prior Work Technology Results 6

Computer Science For All 7

Diverse Majors … with Rich Knowledge English Biological Sciences Animal Sciences Education Chemistry Theater Arts History Building Construction 8

(1) No Prior Background “I’ve never done this before.” 9

(2) Low Self-efficacy “I have no idea how to do this!” 10

(3) Unclear on Why “Why am I doing this?” 11

MUSIC Model of Academic Motivation Students are more motivated when they perceive that: they are eMpowered , 1. the content is Useful to their goals, 2. they can be Successful , 3. they are Interested , and 4. they feel Cared for by others in the learning environment 5. B. D. Jones. Motivating students to engage in learning: The MUSIC model of academic motivation. International Journal of Teaching and Learning in Higher Education, 21(2):272 – 285, 2009. 12

Motivation  Engagement eMpowerment Persistence Usefulness Proactivity Engagement Motivation Success Attendance Outcomes Interest Learning Caring … 13

Situated Learning Beginner Learning • Lave and Wenger • “Learning occurs as a function of the activity, context, and culture” Expert Community of Practice Culture Context Periphery of Community 14

A spectrum FOR-EACH Websites WHILE Games Recursion Iteration Development Mobile Apps IF Algorithms Images Media Content Context Assignment Computation Audio Data Structures Scientific Animations Dictionaries Scientific Modelling Lists Arrays Computing Math Integers Booleans 15

Interesting Contexts 16

Authenticity • Situated Learning • “Relevant”, “Real - world” • Media Computation as an “ Imagineered Authentic Experience” * Mark Guzdial and Allison Elliott Tew. 2006. Imagineering inauthentic legitimate peripheral participation: an instructional design approach for motivating computing education. In Proceedings of the second international workshop on Computing education research (ICER '06). New York, NY, USA, 51-58 17

Why are we teaching computing? “A Tidal Wave of Data” 18

Highlighted Literature • DePasquale 2006 – Real-world web APIs in CS2 • Sullivan 2013 – Data Science for non-majors • Silva 2014 – Big Data in introductory computing • Hall-Holt 2014 – Statistics in introductory computing • Anderson 2014 – Real world data in CS1 • Subramanian 2014 – Visualization of data structures with real data (BRIDGES) 19

Problem – We Need Data • ICPSR – Tightly controlled datasets • UCI Machine Learning – Only for machine learning • Census.gov, Kaggle, etc. – Not ready for beginners 20

Technology • RealTimeWeb – real-time data for introductory computing • CORGIS – real-world data for introductory computing 21

VT Bus Tracking API Dr. Cliff Shaffer Dr. Eli Tilevich

RealTimeWeb – Real-time data 23

So many Points of Failure! U.S. Geological Survey, 2013, Earthquakes Hazards Program available on the World Wide Web, accessed [October 7, 2013], at URL [http://earthquake.usgs.gov/].

RealTimeWeb – Secret Sauce Client Library Online .getData() Web Service [.searchBusinesses()] [.getEarthquakes()] [.getBuses()] [...] Local Cache File 25

RealTimeWeb - Deployment Semester School Course Spring 2013 Virginia Tech CS-2 University of Delaware CS-1 Fall 2013 Virginia Tech CS-2 Virginia Tech Data Structures & Algos Spring 2014 Virginia Tech CS-2 26

RealTimeWeb - Studies N=370, 14% female University of Delaware,VirginiaTech CS1, CS2, and DSA 27

RealTimeWeb - Hazards • Limited APIs • Maintenance was hard • Impact on CS motivation was minimal 28

The Collection Of Really Great, Interesting, Situated Datasets 29

Metrics 44 datasets 267 mB 420,672 rows 9,365,520 values 30

Datasets 31

Connecting to Students’ Majors English Books Criminal Justice Crime Geological Science Weather Aerospace Education Airlines Theater Education Arts History Building Theater Construction Immigration 32 Construction

Architecture Manual Automatic 33

Gallery 34

Java, Python, Racket // Java import corgis.crime.StateCrimeLibrary; import corgis.crime.domain.Report; import java.util.ArrayList; public class Main { public static void main(String[] args) { StateCrimeLibrary scl = new StateCrimeLibrary(); ArrayList<Report> reports = scl.getAll(); } } ; Racket # Python (require crime) import crime (define reports (crime-get-all)) crime_reports = crime.get_all() 35

BlockPy 36

Visualizer Demo 37

Interventions • Computational Thinking Course ❖ Basic programming ❖ Social Impacts ❖ Data Science • 6 semesters taught • Audience ❖ Non-computing majors ❖ Freshmen -> Senior ❖ Gender balanced 38

Course Evaluation • Retention • More-Computing • Gender • Learning Mark Guzdial. 2013. Exploring hypotheses about media computation. In Proceedings of the ninth annual international ACM conference on International computing education research (ICER '13). 39

Survey Timeline 40

Motivation × Course Components Likert Motivational Components Course Component Strongly Disagree “I believe that I will have freedom to eMpowerment "... learn to write computer Programming explore my own interests when I…” programs" Content Disagree "... learn to work with Abstraction “I believe it will be useful to my long - Usefulness Somewhat abstraction" Content term career goals to…” Disagree "... learn about the social Social Ethics “I believe I will be successful in this Success Neither Agree nor impacts of computing" Content course when I…” Disagree Somewhat Agree "... work with real-world data Data Science “I believe it will be interesting to…” Interest related to my major" Context Agree "... work with my cohort" Collaboration “I believe that my instuctors and Caring Facilitation Strongly Agree peers will care about me when I…” 41

Context is Useful N = 85, 62% Female Students’ sense of the usefulness of various course components was highest for the context , lowest for the content . 42

V-Shaped Empowerment N = 85, 62% Female Students’ sense of agency decreases during the BlockPy and Spyder portions of the course, then increases during the final projects. 43

V-Shaped Interest N = 85, 62% Female Students’ interest decreases during the BlockPy and Spyder portions of the course, then increases during the final projects. 44