Increase Enrollment at Becker College with Predictive Analytics
•
1 like•539 views
Predictive modeling has gained popularity in studying college enrollment due to fierce competition in higher education. To make informed decisions and allocate limited resources to improve enrollment, predictive modeling has been applied to challenge and change the traditional recruitment process. This session is intended for two learning outcomes: -Participants who are not familiar with predictive modeling will learn how to lay out a plan to collect and build a comprehensive data infrastructure and conduct predictive modeling. -Participants who have run predictive modeling will learn how to critically examine the quality of their predictive analyses.
Recommended
More Related Content
Similar to Increase Enrollment at Becker College with Predictive Analytics
Similar to Increase Enrollment at Becker College with Predictive Analytics (20)
Recently uploaded
Recently uploaded (20)
Increase Enrollment at Becker College with Predictive Analytics
- 1. BECKER COLLEGE Enhancing the Quality of Predic4ve Modeling on College Enrollment Feyzi R. Bagirov
- 3. Agenda • Enrollment in the US • Background • Predic4ve Analy4cs Workflow – Business Understanding – Data Understanding – Data Prepara4on – Modeling – Evalua4on – Deployment • Next Steps • Take Away Messages • Q&A
- 4. • Founding Director of a Bachelor of Science in Data Science program at Becker College • Faculty of Analy4cs at Harrisburg University of Science and Technology • CBO at 529 LLC (Educa4onal infrastructure and analy4cs) • Data Science Advisor at Metadata.io (ABM, B2B Demand Genera4on)
- 6. Enrollment in the US • The 10-year average for college closures is five annually. • The main struggle for many small colleges is declining enrollment • Moody’s report predicts that inability of small colleges to increase revenue will result in triple the number of closures and double the number of mergers in the coming years.
- 9. “Skunk” Team • Vice President of Enrollment Management • Chief Informa4on Officer • Director of Ins4tu4onal Research • Dean of Admissions • Director of Financial Aid • Director of Data Science • Director of Enterprise Applica4ons • Data Engineer
- 13. Start with a high-level idea • Who is my customer? • What is making my customer complain so much?
- 15. Enrollment Management Process Process: • Prospect/Inquiry Genera4on • Applicant Management • Deposit/Confirma4on Management • Event Management • Marke4ng • Registrar Goals: • Maintain or increase class size • Increase ethnic diversity • Improve academic profile • Increase net tui4on revenue • Lower the tui4on discount rate • Strengthen weak academic programs • Maximize the return of strong academic programs • Support athle4c or other specialized programs on campus Hossler, D., & Bontrager, B. (2015). Handbook of strategic enrollment management. San Francisco, CA: Jossey-Bass.
- 19. Use cases studied prior to design • A major university in the south of the US • Started experimen4ng with analy4cs in 2003 • Used all available data in modeling • Leveraged Clearing House data to see where they were losing students; this allowed to have a be]er idea of true compe4tors. This led to a completely restructured understanding of their compe44on. • Used ACT datasets for profiling
- 21. Use cases studied prior to design • Determining target markets, the Admission’s business model changed dras4cally: – Hiring regional recruiters in areas of high prospect concentra4on – Changing target prospects (high-probability prospects require less 4me, more efforts on 60-80% probability prospects) – Changed the targe4ng process. Customize targe4ng messages. – Changed the types of recruiters they’ve hired – the ones that be]er understand data and use that data to bring in the prospects – Hired a campaign director to track the success of the campaigns – Used in-house callers to spend a lot of 4me on calling campaigns talking to students – Overlaid prospec4ve students with alumni data sets and reached out to alumni to host lunches in areas outside of the home state to engage prospects.
- 22. Use cases studied prior to design • Outcomes: – Launched predic4ve models in 2003. Most informa4on was in Excel. Over4me, acquired a CRM system and leveraged that for consolida4on and accrual of applicant and event data. – Redundant and irrelevant data in CRM was removed over 4me (for example, all students wanted a scholarship, so they did not include that in their model) – Enrollment increase from 20k to 35k over a targeted period, predominantly due to out-of-state students
- 26. Ini4al Ques4ons for Becker • Determining data sources: – ACT, what data elements are included and can be collected – SAT, what data elements are included and can be collected – Clearing House, how long will it take to perform valida4on of students’ enrollment to Becker – Accessing popula4on data to be used in targe4ng areas – Do we know who are our true compe4tors? – What are we doing to target students differently?
- 28. Building In-house Data – IR & IT Work Together, Happily Miller, W. (2016). Conflict Resolu4on. The process took 75% of time of the whole project!
- 33. Predic4ve Modeling Details • Two years of historical data • Descrip4ve: Characteris4cs and demographics (gender, race/ethnicity, geography, first- gen, Pell grant receiver, etc.) • Financial aid: Need-based, merit-based, other grant, loan, work-study, etc. • Behavioral: Applica4on date; deposited date; Admission ac4vi4es (phone call, email, campus visits, etc.) • Logis4c regression • Predict “deposit” probability for each student (values from 0 to 1) • Test each predictor separately first vs putng everything in the model • R/RStudio • Learning curve, but it’s worth Data Model Tools
- 34. • Descrip4ve (Characteris4cs and demographics) – Age – gender – race/ethnicity – Geography (regional – New England/Non-New England, MA/Non-MA, Worcester/Non-Worcester) – first genera4on (parental educa4on) – Pell grant receiver – Sports ac4vi4es count – External ac4vi4es count Data
- 36. • Behavioral: – Applica4on date/applied term – deposited date – Admission ac4vi4es • phone call count • email • campus visits (Acceptance Student Day, campus tours, etc.) Data Admi]ed Accepted Deposited
- 37. • Over 800 lines of code • Majority of code is data transforma4on and cleanup • Supervised modeling • Logis4c regression Model
- 38. Why R? • Open-source (free to use, improve on, and redistribute) • Runs on most standard OS • Released frequently • Graphics capabili4es are be]er than in the most other analy4cal packages • Huge and very ac4ve user community • It provides analysts more control over what changes to make and what assump4ons to test. Tools
- 39. • Excel • RDBMS database • CRM Tools
- 40. Why In-house? • Why in-house? • Most data cleaning can only be done in-house. • Colleges have a be]er control of the modeling process. • The process can be repeated once the codes are wri]en. • More transparency and evalua4on can be done in house.
- 42. Evalua4on step 1 – score models • Is model good enough? – Likelihood Ra4o test – Area under the ROC curve • Are predic4ons accurate? – Confusion matrix • Which Predictors are most important?
- 47. Actually using your model! • Automa4on • Getng feedback from model’s output users • Experimen4ng and feeding the new learning back to the model to improve its accuracy • Monitoring outcomes of the model use
- 48. Deploy Results— Be Responsive & Improve the Admissions Prac4ce 1. The admissions office will receive a file every two weeks. 2. The file will be uploaded to Recruiter. 3. The VP of Enrollment will take ac4ons.
- 49. Challenges in Using Predic4ve Analy4cs • Obstacles in management No champion for the work • Obstacles with data Data need à Data infrastructure à Data consistency • Obstacles with modeling Analyst too zealous or too ambi4ous Model is too complex (overfitng – significant rela4onshpis are just noise)
- 51. Next Steps Build more analy4c work • What areas or school districts should we spend more resources on (visits, calls, marke4ng campaign)? • How do we improve the academic profile of incoming students? • How can we create new aid models to leverage our intui4onal aid to achieve higher yield? • Reten4on - which students are most likely to drop out/transfer? • U4lizing unstructured data to gain addi4onal insights Actually USING the results • What ac4ons need to take place based on scores generated from predic4ve models? • How do we assess the effec4veness of our upliuing marke4ng campaigns (including customized emails, number of phone calls, and other marke4ng ac4vi4es)?
- 52. How else can data analy4cs help my school? • Increasing students’ reten4on • Increasing students’ gradua4on • Increasing students’/teachers’ performance • Reducing students’ absence
- 53. Take Away Messages • Be realis4c to narrow the scope of the predica4ve modeling work at the ini4al stage • Evalua4on of models should be in the process • Building data structure is more important than methodology