From Data Aware to Data Driven Our Journey

From Data Aware to Data Driven: Our Journey Towards Predictive Analytics Phebe Soliman, Director, Institutional Research Michael Arabitg, Business Intelligence Analyst County College of Morris #RIConf 2019

From Data Aware to Data Driven: Our Journey Towards Predictive Analytics Rapid Insight User Conference 2019 Phebe Soliman & Michael Arabitg CCM

Introductions • Phebe Soliman – Dean of Institutional Research • Over 15 years higher education institutional research experience • Areas of expertise are in data structures and governance, data reporting collection & use, and survey design • Michael Arabitg – Business Intelligence Analyst • 3+ years with County College of Morris • IR Technician from 2014 – 2016 • Maintained analyzed birth certificate database at NJ maternal and child health consortium

Overview of CCM • The college was established in 1968 • The main campus is located in Randolph, NJ on 222 acres. • We are a 2 year community college and offer 58 associates degrees and 22 certificates • Enrollment for Fall 2018 was approx. 7, 500 students • 1, 400 sections taught • 76, 000 credit hours

Agenda • CCM’s past, present, and future environment • How CCM introduced predictive analytics to the college community • Why we selected Veera • How we incorporated it • Communication to stakeholders • Discuss the first projects to identify students likely to benefit from outreach • Closing/Questions

How do you spend your time? • What proportion of time is spent on: • • Regulatory and compliance reporting? Internal or external custom, ad-hoc requests? Repetitive, manual data assembly? Strategic / research data gathering and analysis? • What activities or requests are left unfilled?

The Past

The Past • All data was retrieved from Colleague (SIS) via Operational Data Store (ODS), overwritten daily. • • • There was not an easy method to capture point-in-time comparisons. Any longitudinal or historical analysis was cumbersome to create. Recurring reports would take days, if not weeks, to generate. Could not track student movement or changes within the term easily and effectively. Tenth day (census) reporting was tedious.

Data Maturity Model

The How CCM Introduced Predictive Analytics to the College Community

From data aware to data driven How CCM introduced predictive analytics to the college community • Started with support from leadership • Started with the end in mind – no matter the tool, if the data isn’t structured correctly the tool won’t work • IR & IS created a small data warehouse, built in-house that captured select student and course information. The data was run and captured daily. Started as a side project to alleviate federal reporting • Built data warehouse – Student Universe • Establish Data Governance Council • Subscribe to Rapid Insight – predictive analytics

CCM – System Overview Colleague is our ERP: • • • Finance, Human Resources, Student, CRM Recruit, Time Reporting, Student Planning and Self Service Windows Environment SQL Server 2012 Blackboard is our LMS Data Warehouse (internal) Reporting: Business Objects version 4. 2 SP 5 Patch 4 SQL Server Report Services (SSRS) SPSS, Excel and other tools

Reorganization at CCM • New Leadership – President began 2016 with a data driven focus • Merger of Institutional Research and Information Systems in 2017 • New position VP of Institutional Effectiveness/CIO • IR consists of 3 people t n a ac e v n (1) Dean Institutional Research, (2) Business Intelligence Analysts O • IS consists of 5 people (1) Associate Director of Information Systems, (1) Colleague Systems Administrator, (2) Programmer Analysts, (1) Senior Systems Analyst

The Driver: Student Success • Achieving the Dream (ATD) • Need to respond to leading as opposed to lagging indicators. • New leadership – came from institution where data was heavily used • Understands the significance of accurate data • Hold the philosophy that everyone must own their numbers

Data Warehouse for Business Intelligence • • Worked with ASR Analytics (a consulting company) Helped us develop a comprehensive longitudinal data platform Required focus – high priority Leveraged IR/IS data mart as starting point Pause on data requests (sorta) Data governance Training – external and internal Supported by leadership! (can’t emphasis that enough)

Data Governance Council (Affectionately known as DGC) • In an effort to integrate CCM’s separate data systems into an enterprise view of data the establishment of a Data Governance Council is essential to ensure the data quality. Purpose of the Data Governance Council: • Create a single source from which to present accurate and reliable data to all who use it within the institution, and, where appropriate, to customers and other consumers of data external to the institution. • Identify, establish and oversee the strategy, objectives, rules, procedures, practices and technology ensuring that quality data is provided to the institution.

Data Governance Council -Function. DGC Function: just a few • Establish and maintain a data definition dictionary and coding standards for the college’s critical external compliance and internal operations reporting requirements for use by the institution • Identify, establish and oversee processes for data corrections at the source or data‐entry level that are based on established data definitions and standards (ex: exception reports, data correction parameters) • Provide continuous training on data standards, metrics and usage to the responsible data owners and the college community

Data Governance Council -Membership. Cross-functional team • Chaired by Senior System Analyst and Dean of Institutional Research • Membership is based on the College’s primary data entry point offices, external and internal reporting specialists from those areas that are responsible for compliance reporting • Includes experts in the student system, academic affairs, institutional research, information systems, human resources, finance, financial aid, and vendors (as necessary) • Standing members – Appointed as a job function • Ex‐officio/Ad‐Hoc members – Appointed to fill a process or input specific function

DW/BI process Now that we have the right people and tools – what did we do with it? • • Gauged input from administration and DGC on the types of reports Establish 6 global drill down reports- allow for hands on approach Built new reports for more focused reports Training – ongoing

Daily Enrollment Report

Example

High Level View

Training • • Lost of reports = lots of (continuous) training Power users and end users Training schedule Different focus for different groups

The Present

The Present • Still in infancy within college community, however it is gaining momentum and cabinet is onboard. • Deans are beginning to use standard reports for enrollment and retention management. • Developing new report request process. • Looking at ways to improve delivery and appeal. • RI –Predictive analytics

Using Rapid Insight – Veera Attrition vs Retention • Model is built for attrition (non-continuation) • Built it for action (identifies characteristics of students not likely return) • Sharing student list with Academic Success Center • Jobs can be automatically scheduled to run based on different models to generate a list of at-risk students • List can be emailed to advisors to contact these students

Early Output Average attrition rate

Early Output

Using Rapid Insight – Veera DFW Analysis • Identified courses with highest DFW rates among students enrolled • Mostly Math & English, specifically developmental • Criteria: 3 -year averages, 500 grade minimum, W = 0, location specific • Results were shared with Deans, Chairs, and Faculty

Data Maturity Model

The Future

Future Data Growth • Implementing admissions data • Prospects activity • Adding LMS data • Attendance, grades, student activity • Review other data sources Continuously reviewing and learning to make better!

Future Analytics • • Better, faster reporting (one-click) Cleaner data i. Data Cookbook Predictive modeling • New models: student program changes, course taking patterns, and enrollment projections • Cloud-based reporting Review entire student lifecycle from admissions to graduation and beyond to increase student outcomes and success

Data Maturity Model Then Now Future

Thank you! Questions/Comments? Phebe Soliman, Dean, Institutional Research psoliman@ccm. edu Michael Arabitg, Business Intelligence Analyst marabitg@ccm. edu County College of Morris #RIConf 2019