1. Program Review Support Tool Nathan Pellegrin Research Analyst

2. Goals Background and purpose of the tool Demonstration Cal-PASS update OLAP Development at Cal-PASS OLAP Success Story: SSPIRE Cube Future development

3. The Program Review Support Tool Funded by Hewlett. Currently being tested by several colleges. All data is from MIS. Caveat: figures may not match what is found in locally produced reports due to differences in master data sources and formulae used to derive figures. Like a “data smorgasbord” and includes –student demographics –course grades –TOP code course hierarchy … the menu will be expanding !

4. Purpose Not the product of a mandate or requirement from the Chancellors office. Not intended to take the place of local tools. Not intended to drive evaluation activities. Intended for use by colleges as an optional FREE tool in their program review process. Obtain feedback from users to scale and improve our data model and OLAP infrastructure.

5. Cal-PASS Statistics  Over 300,000,000 records  Up to 15 years of data in some regions  Over 7,000 schools, colleges and university members  Over 150 research studies conducted in the last two years  Sixty-six Professional Learning Councils (1,200+ faculty)

6. Universities (23) CSU. Channel Islands Dominguez Hills Fresno Long Beach Los Angeles Monterey Bay Pomona Sacramento San Bernardino San Marcos Stanislaus San Francisco San Jose Sonoma UC. Davis Merced Riverside San Diego Santa Barbara Santa Cruz PRIVATE. Otis College of Art and Design National University University of the Pacific

7. Changing the Paradigm: OLAP Applications OLAP = On-Line Analytical Processing Like Excel pivot tables, except Excel handles only two dimensional data. Stores pre-computed aggregations of data with B-Tree indexing for delivering fast retrieval times and fast calculation. Enables users to perform analysis of data quickly with drag- and-drop manipulation of variables and dynamic visualization. Web-based for easy access – all processing is performed on the server so it does not tie up your work station (zero footprint). Big time savings! Ideal for the action research paradigm and design research.

8. User Interface - Dundas OLAP Cube - SSAS Database(s) – SQL Server MDX SQL 3 Layers of the Application

9. Development Process of the OLAP project is a technical collaboration between IT and Research … Server Architecture/O.R. – Alex Zakharenkov (IT) Submission Processes/User Interface – Nick Wade (IT) Data Model/ETL – Nathan Pellegrin (Research) Design/Feedback of OLAP cubes - All IT and Research Staff, including Terrence Willett and Mary Kay Patton

10. Development times Development of initial Dim Model started in July 2008 … incremental additions/changes congealed into a (basic) model by February, 2009. Initial development of Program Review, including feedback and changes ≈ 8 weeks. Dim model ETL execution ≈ 15 hours. Processing of OLAP cube ≈ 20 min./300K rows. Initial deployment of UI ≈ 3 weeks. Several changes since then. UI required tweaks to OLAP cube design.

11. Development Tools.NET SQL Server –storage –Integration Services –Analysis Services BIDS Dundas

12. LA Object Repository UC CCCCO MIS K-12 UC CSU Private CSU Universities Cal-PASS Submission ETL 0607,01612590000001,0000179441,U,,12,ABARCA,CARL OS,09091988,M,500,,,,,,,,,,,,,,,,,,,01,15,1,,,,,,N,010,Y,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,01,10032006,N,275,0,0,4,24,0,0,0,0,0,0,,1 0032006,X,,,,,,,,,,,,,,,,,,,,,,,N,Y,,,8,,,,,,,,,, 0607,01612590000001,0000154281,9107510861,,11,BLAC K,BRITNI,11291990,F,600,,,,,,,,,,,,,,,,,,,00,13,1,,,,,,N,000,N,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,01,10032006,N,302,1,8,6,35,3,15,5,28,4,33,01,10032006,N,340,5,71,12,67,18,90,0,0,7,47,2.5,,,,,,,,,,,N,N,U,72,80,,,,,,,,,, 0607,01612590000001,0000159553,U,,11,BOWIE,EARLIS HA,10231988,F,999,,,,,,,,,,,,,,,,,,,00,14,1,,,,,,Y,060,Y,,Y,,,,,,,,,,,,,,,,,,,,,,,Y,,,,,,,,,,,,,,,,,,,01,10032006,N,278,3,23,4,24,4,20,0,0, 0,0,,10032006,A,,,,,,,,,,,,,,,,,,,,,,,N,N,,,40,,,,,,,,,, 0607,01612590000001,0000161233,U,,        9107510861,,11,BLACK,BRITNI,11291990,F,600,,,,,,,,,,,,,,,,,,,00,13,1,,,,,,N,000,N,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,01,1003200 6,N,302,1,8,6,35,3,15,5,28,4,33,01,10032006,N,340,5,71,12, 67,18,90,0,0,7,47,2.5,,,,,,,,,,,N,N,U,72,80,,,,,,,,,, 0607,01612590000001,0000159553,U,,11,BOWIE,EARLIS HA,10231988,F,999,,,,,,,,,,,,,,,,,,,00,14,1,,,,,,Y,060,Y,,Y,,,,,,,,,,,,,,,,,,,,,,,Y,,,,,,,,,,,,,,,,,,,01,10032006,N,278,3,23,4,24,4,20,0,0, 0,0,,10032006,A,,,,,,,,,,,,,,,,,,,,,,,N,N,,,40,,,,,,,,,, 0607,01612590000001,0000161233,U,, CUSTOM FILES Analytical Integrated K-12/CC/Univ Time-dependent 2NF (Redundant CK) Optimized Indexing ETL Dimensional Model Semistructured data Format/value Validation Storage Application Integration Key-value pairs (KVP) design Cal-PASS Data Flow

13. What does a dimensional data model do for Cal-PASS? UNIFY: Data from across segments is integrated into a unified dataset. STANDARDIZE: Table and field names, data types and value coding systems are standardized to be the same for all segments. SIMPLIFY: The number of tables and fields used to store the data is reduced. Granularity of tables are at the units of analysis. Table relationships reflect analytical relationships between entities. BOOST PRODUCTIVITY: The simpler, cleaner data model makes it easier to develop cubes with re-usable components, generalized for all segments. Currently, analytical data processing must be developed separately for each segment. Using a dimensional model, only one pathway needs to be developed that applies to all segments. IMPROVE DATA QUALITY: Merging data brings data quality issues to light so they can be noted and/or resolved. Establishing primary and foreign key relationships enforces referential integrity. Multiple student identifiers are unified to produce a single “metakey” Missing course CBEDS classifications imputed using machine learning. REDUCE RISK: Without it, in order to produce one metric for all segments separate analytical data processing pathways are required for each segment, which means more maintenance and increased risk of inconsistent results. Using a dimensional model the analytical computations and services are centralized.

14. Organization Student Course Term Course Taxonomy Student Status Course Outcome Award Cal-PASS Unified Dimensional Data Model (Selected Tables) = Fact Table = Dimension Table = Foreign Key Relationship

15. Dimensional Model Tables Dimensional Model Tables The Ideal: Centralization of analytical query processing Each statistic can emerge at multiple presentation points, but there is only one logical control point. Views and Stored Proc’s OLAP Presentation & User Engagement User-defined cohorts; model outputs

18. Student identifiers from each source system are mapped to a new identifier through transitive closure of all connected values (using a modified version of the Floyd–Warshall algorithm). Local district student id CSIS SSID Name + gender + DOB CCCCO SID (SB00) n1 d1 d2 n2 o1 c1 d3 m1 Each edge represents a record linking two values of different identifiers in submitted student records.

19. OLAP Success Story: SSPIRE Cube Funded by Irvine Foundation. Currently used by nine colleges. Incorporates MIS data with data submitted by colleges (custom files). Tracks cohorts of students. Demonstrate using Merced college (thank you Dr. Duran!)

20. Program Review Support Tool

21. This is only the beginning… Provide access to K-12 districts and Universities Inter-segmental OLAP Cubes Link non-academic outcomes (Employment Development Department, Child Welfare Data System) “Success at Every Level”

22. Education Data and Information Act of 2008 SB 1298 1. convene a high-level working group to decide the best the governance structure for the comprehensive education data system; 2. directs the State Chief Information Office (CIO), in consultation with educators and education policymakers, to prepare a strategy plan outlining a clear path for technical implementation; and 3. requires the various education segments to begin using a common student identifier, so that once a governance structure and technical architecture are in place we can begin linking records from pre-k through the university with relative ease and speed. Source: http://www.senatorsimitian.com/legislation/entry/sb_1298_education_information_system/

23. 23 CDE Data Systems High level cross-agency systems map of key collections WORKING DRAFT NOT EXHAUSTIVE CASAS TOPS Pro SACS Assessments CALPADS** CASEMIS Migrant ConApps AYP/API Early Childcare CALTIDES** CPEC CALPASS UC CSS EDD CSU ERSCCC COMIS CCTC CASE Other CDE systems/ units including CDS, Charter schools Other CDE units including Homeless, CALSAFE, Title 3, Private Schools etc. Non CDE Data Systems Prisons, Census From Franchise tax, benefits system etc. National Student Clearinghouse Data sharing through local agencies Direct data sharing* Planned/ potential * Does not imply direct data linkages. Only state system linkages shown ** CALPADS is envisioned to replace much of the CBEDS, Language Census, Student National Origin Report and select Consolidated Application data *** CALTIDES is envisioned to collect data primarily from CALPADS and Commission on Teacher Credentialing’s CCTC’s Credential Automation System Enterprise CASE system Source: Interviews with respective agencies, RAND, team analysis In development Existing CDPH Source: http://www.senatorsimitian.com/legislation/entry/sb_1298_education_information_system/

24. 24 High level system profiles of key CDE collections (1/2) System nameDescription CALPADSCalifornia Longitudinal Pupil Achievement Data System. System (under development) for tracking K12 students longitudinally, that will replace CBEDS collections Key identifierData categoriesGranularityData sharing SSIDStudent demographic, program participation, grade level, enrollment, course enrollment and completion, discipline, and statewide assessment StudentPlanned include- Assessments, API/AYP, Migrant, ConApps, CALTIDESCalifornia Longitudinal Teacher Integrated Data Education System. Iintegrated data system for teacher data based on unique SEID SEIDTeacher credentials, authorizations, teacher participation program, alternative routes, participation in Beginning Teacher Support and intern program, SEID, Salary StudentPlanned include- CALPADS, CCTC CASE CASEMISCalifornia Special Education Management Information System. Integrated data system for special education students on students, services and provider programs SSIDAttendance/Enrollment, Disciplinary, Education Agency, Mobility, Special Education, Staffing Data, Student Demographic, Other (services, age, gender, race/ethnicity) Student, School district, School, county, region None at state level AssessmentsCalifornia High School Exit Exam CAHSEE, Standardized Testing and Reporting STAR and CELDT SSIDAttendance/Enrollment, Education Agency, Food and Nutrition, Parent Data, Special Education, Student Demographic Student, School District, School, County CASAS, Migrant, AYP/API, CALPADS (planned) API/ AYPAccountability related information based on California's Public Schools Accountability Act of 1999 as well as No Child Left Behind Act of 2001 CDS codeAYP/API score by student characteristicsSchoolAssessments DRAFT Source: Respective CDE departments Source: http://www.senatorsimitian.com/legislation/entry/sb_1298_education_information_system/

25. 25 High level system profiles of key CDE collections (2/2) System nameDescription MigrantStudent enrollments in migrant education programs. Includes migrant education forms and a directory of offices providing services Key identifierData categoriesGranularityData sharing Migrant ID, COE number, CDS code Student demographics, educational programs, counseling, health and support services, emergency health, clothing, food, transportation StudentAssessments, CALPADS (planned) SACSStandardized Account Code Structure. Offers LEAs with a means of reporting financial information CDS codeFor every general ledger accounting transaction- information on funds, resources, project year, goal, function, and object. Includes information on Attendance/Enrollment, Education Agency, Fiscal, Transportation School, DistrictCDS, Charter schools ConAPPSConsolidated applications. Includes information on categorical programs e.g., Title I, II, V etc. CDS codeStudent demographic, Title I, III, V, Part A, Immigrant, LEP, funding model, charter status, Gradespan, participants School, District, County CALPADS (planned) Early Childcare Systems CD-801A,B, CDMIS, Special Education Desired Result System SEDRS, and CD 9600 SSIDChild demographics, IEP flag, family identification/case number, household name, type of program, DRDP Desired Result Development Profile, Early Childhood Environment Rating Scale ECERS StudentNone CASAS TOPSPro Comprehensive Adult Student Assessment Systems. System for tracking Students in Adult Education Programs ADA ID, SSID, CASAS no Student demographics, Agency, instruction level and program, assessment scores, date of entry, reason for exit, class number, attainable goal within program year StudentAssessments DRAFT Source: Respective CDE departments Source: http://www.senatorsimitian.com/legislation/entry/sb_1298_education_information_system/

26. 26 High level system profiles of non-CDE collections System nameDescription CPECCalifornia Post Secondary Education Commission. Data system for Higher Ed- post secondary systems Key identifierData categoriesGranularityData sharing Student ID based of SSN Demographic, IEP, grade level, program, Graduation rate, teacher, institution StudentCDE, CSU, UC,CCC, prison, census UC CSSCorporate Student System provides information on student enrollment and performance for University of California campuses SSNStudent demographic, income, financial aid, education history, assessment StudentCDE, CCC, CALPASS DRAFT CCC COMISCalifornia Community Colleges Management Information System. COMIS data is used to prepare reports for Federal and State reports including Integrated Postsecondary Education Data System (IPEDS) and to track student outcomes SSN, Student ID student demographic, income, financial aid, education history, assessment, teacher, institution StudentCALPASS, CPEC, CSU, EDD, National Student Clearinghouse CSU ERSEnrollment Recording System is used by Cal State to track student retention and graduation to support regular term reports, IPEDS, and state budget requests SSNStudent demographic, financial aid, education history, assessment StudentCPEC, CALPASS, CCC CDPHCalifornia Department of Public Health. System use to track CDPH IDCase ID and demographics, clinical and diagnostic data CaseNone EDDEmployment Development DatabaseID based of SSN Wages, payroll taxes, unemployment tracking, job matching, job training EmployeeFranchise tax, benefits system, CCC Source: Respective agencies, RAND Source: http://www.senatorsimitian.com/legislation/entry/sb_1298_education_information_system/

27. HAPPY DATA! Data Thank you! Have fun ….