Lech Maj (NYU)
Becky Joffrey (Cornell) –
Student success – the ability to have a 360 degree of the student lifecycle
How involved is IT team with student success initiatives – 66 (out of 100)
How involved is IT team with learning analytics: 70 (out of 100)
NYU – Bernie Savarese (Assistant VP for Student Success)
Reports through Enrollment Management (which includes Peoplesoft, Fiscal aspects, and Recruitment)
Student Success is everyone’s business!
Why? Position & Rank; Financial Stability; Perceived Value & Alumni Engagement; Deliver on Promises
First year retention and Graduation rate make up almost 1/3 of US News ranking. Found NYU was lagging peers in those measures.
Financial Loss – 1st year attrition Net loss $16 million
Goals: FIrst year retention – 96% by 2020, Six year graduation 90% by 2026: require keeping 50 extra students per year.
Student Success Steering Committee, with a Technology Task Force.
Guiding Principles: Use technology to drive relationships; Make a big complex place feel small: find and support students who need us most; identify and remove unintended barriers; find and surface evidence for continuous improvement.
Chose Starfish as a technology platform – launching in Fall 2018. Why a platform? See the whole student, aggregate critical information and systems, coordinate care, simplify resource referral, identify leading vs lagging indicators, deliver on promises.
Platform goals for Year One: All undergraduates in all schools; raise flags and alerts; predictive analytics and risk scores; appointment scheduling; shared notes; include student affairs/services
Need to be able to close loops – e.g. if a faculty member raises a concern about a student, they need to know what happened.
Students have the ability to see what’s going on in a dashboard.
Becky Joffrey – Cornell’s Student Engagement Ecosystem
Current State: Role based infrastructure (point solutions with a single role (club member, dorm resident, job seeker) in each system). Data lives withn each transactional application. Spend a lot of time marrying data from systems, but still struggle.
Desired state: Person-based infrastructure – many roles that change over time, data moves and grows with a person; invest in understanding constituents. Duh – CRM.
Ended up with 23 Salesforce orgs – that’s no way to implement CRM. Caused Provost to start project to think about student experience globally. Move from departmental intent to institutional intent.
Led to strong steering structure and governance. Provost funded, steering committee led by a dean and the Vice Provost of academic affairs, experience working group, analytics working group.
Initiative: Modernize technology to support student experience; focus on student services advising student activities and analytics. Goal is to connect all parts of the student experience. Audiences include students, the people at Cornell who support them.
experience.cornell.edu – Discovery website for student opportunities. When you click to apply for an experience, it takes you into Salesforce. A rich dashboard experience for the student that integrates and orchestrates all the different experiences. Advisors also have a dashboard. Other web sites can use the information to display filtered views of opportunities.
Putting finding tools in Drupal sites (Opportunities, Resources, Clubs, Events, People), and “Doing Things” in Salesforce (transactions)
The benefit is data: You can see who is doing X, but more importantly you can see who is NOT doing X; Data is collected via natural points of engagement vs. surveys and notes; Data benefits the entire institution, not just individual unit. Prioritize apps that will glean the richest sets of data.
Tips: Find a point of gravity that brings campus together; start with users’ problems; identify urgent need; build horizontal solution, not a vertical one; consider breadth of tools available and how they integrate; create and extendable architecture
UC Berkeley – Oliver Heyer – BOAC and the Data Loch
Early Work: CalCentral – collects data from a variety of sources to allow students to do the work they need to do. Became student front-end to new Peoplesoft SIS.
Cloud LRS (learning record store) in AWS. Pulls in feeds of data for storage and analysis. Built LTI tool (student privacy dashboard) on caliper data from Canvas. Not in production yet.
Athletic Study Center didn’t have a view into Canvas data. Putting advisors into Canvas as observers didn’t provide a manageable way to provide information.
Berkeley is framing student success around issues of diversity and inclusion.
BOAC/Data Loch solves some important problems: Canvas Big Data > UCB Data Lake > Learning Analytics; Custom cohorts; An early warning system; Stability, security, and scalability. Have about 900 students being used by 40 advisers. Storing cached data from APIs in AWS (live use of local APIs didn’t scale)
Goals: Bird’s eye view on learning and other data emerging from varied sources; Data Collection layer; Data processing layer (redshift, spark, athena); Deliver insights as a service (description, predictive, prescriptive)
Using AWS Glue for ETL into data catalog, which can be queried by Redshift Spectrum and then tables extracted into Redshift.
Largest dataset – canvas event logs. 650 GB, ~4.5 billion requests records. Compressed into Parquet. Did a table scan with Redshift – cost $1.25 and took 5 minutes.
Next steps: Expansion of advising to College of Engineering in fall 2018. How to tell story to faculty? EDW collaboration – could move into data lake. Where does application live? Is it yet another place to go? Implications for campus data and cloud platform strategy in general?
MyUW: Supporting the Student Lifecycle – Jim Phelps (Washington)
Used User Center Design including a student diary where they asked students to collect their information needs.
Findings: information overload; critical information hard to find; time management is difficult; information needs are dynamic, but predictable
Design goals: Personal, critical, curated, relevant, timely
Arrive at: actionable and personalized content on “cards”
Show students what they most need based on time in the quarter. e.g. where are my classes in weeks 1 and 2, when can I register in week 5, when are my finals in week 8.
Aggregates data from multiple sources.
Understanding the student lifecycle experience – transitioning to UW, exploring majors, in major, transitioning to profession
Understanding the co-curricular experience – Present interest, social catalyst, internalized motivation, major blocker?, information seeking, participation in co-curricular. How do we build a social component to help students connect?
Husky experience toolkit – tailored messages.
Assessment – continue to assess effectiveness and usability if MyUW and Husky Experience Toolkit. Surveys, log analysis, “guerilla” user studies. Feeds back into user-centered design process.
UC San Diego – Harnessing the power of analytics for student success – Amin Qazi and Christopher Rice
What is a student? Matriculated; non-matriculated; extension; undergraduate/graduate/professional; how long is one considered a student?
What constitutes success? better grades; improved time-to-graduation; retention; quality of experience; getting needed courses; improved job performance or advancement; personal satisfaction
What do students want? High quality degrees that are more career oriented and time compressed; sequential programs linking graduate offerings; dual graduate programs; online degrees from reputable institutions; stackable progression models
Building advanced data capability to: prepare the university for the application of AI and Machine Learning; guide reallocation of scarce resources in data driven ways; harness automation; empower units to harness the power of analytics
Overview of next generation data warehouse – layered architecture. Core data in middle, applications feed data in and out, connected by APIs.
Platform predictive capabilities (builit on SAP Hana) – also working with Google and Amazon.
Bringing all information from university into a single data warehouse. Activity hubs (employee, student, academic activity, facilities, financial activity, advancement & alumni activity). Working on student activity hub first. Real-time data, personalized messaging and interactions, complete data integration, next-generation data science. Three classes of analytics: institutional, academic, learning.
Curated view of data, de-identified – demographics, enrollment, majors/minors, retention, student statistics per term, etc. Reports are generated by people in business units, delivered by Tableau or Cognos or API. Multiple levels of security. 10 years of data from SIS, seven years of data from LMS, rolling out now.
Goal is to have four activity hubs up by end of year, and sunset the enterprise data warehouse in two years.
Retrospective and Predictive analytics for student success.
Architecture is not enough. You must build a culture around analytics: Communities of practice; data governance committee; missionary work; easy to use platforms & tools; pushing analytics to the edges, away from ITS.
Strategic Academic Program Development (use data to build RFPs) – meritocracy of ideas, reach across campus, experiment (fail, learn, repeat), focus on what is best for the student?