1. Sherpa, MAP, Predictive Analytics and Dashboard: Software Designed for Success
Bob Bramucci, Ph.D. Vice Chancellor, Technology and Learning Services
Jim Gaston, Director of IT – Academic Systems

2. About Us About the SOCCCD Two Colleges, Three Campuses
Two-college district in southern California
Approximately 41,000 students (23,000 FTES)
Two Colleges, Three Campuses
Saddleback College in Mission Viejo
Irvine Valley College in Irvine
Advanced Technology and Education Park (ATEP) in Tustin

3. SOCCCD Student Success Systems
MAP
Sherpa
Predictive
Analytics
Student
Dashboard

4. The Problem

6. The Planets Align…

11. Progress So Far

12. Technology & Productivity

13. Moore’s Law

14. Computer Prices

15. Communications Speed

16. Search Time

17. Computers often exhibit Exponential rather than Arithmetic efficiencies

18. Disinvestment by States

19. We Can’t Get There Alone
80-90% spent on salaries + benefits in K-12 and postsecondary education limit options (we can’t realistically double the number of people)
Declining state investment
Economic history shows that technology often produces exponential changes
Is increased automation the answer?

20. SOCCCD’s SECRETS
It’s not just what we created (an integrated suite of student success software applications)
It’s How We Built It.

21. SOCCCD’s Secrets Student-Centered Design
Agile Software Development Methods
Inclusive
Flexible
Human-Centered Allocation of Function

22. Student-Centered Design

23. Agile Software Development

24. Allocation of Function
Vs.

25. Fitts’ List HUMANS Complex pattern recognition and generalization
Judgment under uncertainty
MACHINES
Rote calculation
Routine, repetitive, precise movements

26. Human-Centered Allocation of Function
We believe that simply replacing people with machines is NOT the best answer
Rather, we can allocate functions between humans and machines in a strategic way that allows people to leverage their unique human strengths--a “human- centered design” approach

27. SOCCCD Student Success Systems
MAP
Sherpa
Predictive
Analytics
Student
Dashboard

28. My Academic Plan (MAP)

29. My Academic Plan: Helping Students MAP Their Future
Jan 2009
MAP Project Goals
Create a system that provides:
Self-service guide for students to develop academic plans
Tracking system for counselors and students to monitor progress toward goals
Automated assistance to help students select classes that achieve their goals
Integration with Project ASSIST to reduce data maintenance and increase accuracy
Community College Futures Assembly

30. My Academic Plan: Helping Students MAP Their Future
Jan 2009
Process
Design Team
Counselors
Students
Articulation Staff
Timeline
Discussions began in late 2004
Development began in Fall 2005
MAP went online on April 27, 2007
Technology
Service Oriented Architecture (SOA)
Integration with Project ASSIST
Community College Futures Assembly

31. Usage Statistics - Plans Created
My Academic Plan: Helping Students MAP Their Future
Jan 2009
Usage Statistics - Plans Created
Irvine Valley College
Associates
17,692
Certificate
6,329
Transfer
74,604
Saddleback College
42,729
17,302
102,704
Total:
261,360
From: 4/27/07 to 9/26/14
Community College Futures Assembly

32. My Academic Plan: Helping Students MAP Their Future
Jan 2009
MAP Demonstration
Community College Futures Assembly

33. Sherpa

34. Sherpa So what is Sherpa?
Sherpa is a recommendation engine that guides students to make informed decisions regarding courses, services, tasks and information.

35. Sherpa Architecture Timing Event Trigger Courses Services Information
Tasks
Transcript
GPA
Ed Goal
Reg Activity
Current Classes
Status
Recommendations
Student Profile
Nudge
Feedback - Preferences
Feedback - Ratings
Portal News Feed Text To-Do List Mobile

36. Let’s translate this into the real world…
Nudge
Profile
Here is some advice for you… ?

37. What makes a nudge, a nudge?
Personal
Timely
Relevant
Actionable

38. Recommendation Examples
Student Profile
Recommendation
New student, not matriculated
Link to online orientation
GPA > 3.5
Link to honors program
GPA < 2.0
Link to tutoring program
Transfer student
Courses that fulfill GE pattern
Does not have academic plan
Link to MAP (academic planning tool)
Low assessment scores
Recommend basic skills classes
First time student
Step-by-step instructions
Returning student
Recommend classes based on goals, success patterns and transcript details. Only display classes with an empty seat and that don’t conflict with student’s schedule

39. Phased Approach Phase One – Closed Class Assistance
Assists students in finding a replacement class during registration
Phase Two – Profiles and Nudges
Announcements, , SMS, Voice-To-Text
Phase Three – Portal Refresh
Calendar, To-Do List and News Feed
Phase Four – Mobile
In beta release
Phase Five – Predictive Analytics
In development
Phase Six – Student Success Dashboard

40. Some Promising Early Results
Closed Class Assistant:
Has directly resulted in over 25,400 enrollments since October 2010.
Matriculation Pilot:
Probation numbers dropped 39% one month after Sherpa notification.

41. Results: More Engagement
Yesterday I gave an IVC “Freshman Advantage” presentation to almost 100 high school Seniors at Segerstrom High school. I have been giving these presentations for years. I cannot BEGIN to tell you the positive difference in the student’s behavior due to the Sherpa nudge. It is like they had been touched by an electrified cattle prod in a GREAT way! Students were literally crowded around me before the presentation SHOWING me the Freshman Advantage message on their smart phone!!! They had a higher level of awareness, a feeling of more urgency to complete matriculation and a better understanding of the process. - Irvine Valley College Outreach Specialist

42. Sherpa Demonstration

43. Predictive Analytics

44. Uses
Early Alert
Course Recommendation
Adaptive Learning

45. Machine Learning Decision trees Neural networks Bayesian networks
Learning Automata
Support Vector Machines

46. Three Math Models
Global: is student predicted to fail any of his/her courses?
Specific: A,B,C,D,F grade prediction for any particular class a student might take
Regression Model

47. Data Scientists

48. Problem Structure
Millions of movies—but you’ll only view a small subset
Thousands of courses—but the average student will only take a small subset

49. Matrix Factorization Model

50. Student Success Dashboard

51. Focus on Student Success

52. Next Steps
Research
Scaling Up
Additional success modules

53. Our Roadmap

54. Questions? http://www.socccd.edu/sherpa
Bob Bramucci: Jim Gaston: