A Predictive Model to Help Students Excel After Graduation
The prestigious business school aimed to develop predictive models leveraging years of accrued data and remained open to a wide range of possibilities. Having previously collaborated with the JAKALA team on technology, data management, and business intelligence solutions, the university entrusted us to explore a variety of ideas.
Ultimately, the Graduate Fulfillment Model was chosen, which evaluates both the likelihood of graduates securing a job and their fulfillment with their careers. This university prioritizes student success and career satisfaction, making this model essential to the school’s mission.
The Graduate Fulfillment Model aims to:
- Identify the key variables in students' educational journeys, and also external contextual variables, that impact their likelihood of career fulfillment.
This model allows the university to predict future student satisfaction and focus on improving it throughout the two-year period. With insights from the model, the university can enhance its data assets and analytic capabilities to further support career development and student satisfaction. JAKALA also developed a dashboard that gives stakeholders easy access to student fulfillment forecasts and recommended actions to improve them. Additionally, integrating new data into the model enables continued improvement in student outcomes over the two-year period.
Other potential models JAKALA presented:
- Improving student retention
- Improvement of student performance
- Optimization of course planning
- Identifications of behavioral patterns
- Improving the effectiveness of teaching
- Optimization of admission processes
- Personalization of student experiences
- Forecasting course demand
- Identification of funding opportunities
- Optimization of university operations
- Identification of skills required by the labor market
- Improving the teacher selection process
- Identification of training needs of academic staff
- Forecasting recruitment costs and optimizing marketing activities
- Predicting the academic performance of international students
More Efficient Data Parsing with Generative AI
Generative AI enables efficient parsing of documents such as resumes and certifications in various formats, extracting specific information for streamlined analysis. In the next phase of our project with the university, we applied generative AI to parse student documentation, including certifications and curricula. Using generative AI did not just allow the extraction of simple information already contained in the curriculum vitae (CVs), but also enabled the reprocessing of data to generate new insights and new types of aggregated information from the CVs.
We created an AI-driven process that automatically analyzes and collects data from student resumes and certifications, which integrates seamlessly with the university’s existing data lake. This enriched dataset enhances the model’s accuracy, providing deeper insights into the variables that impact students' post-graduation success and fulfillment.
Primary Goals:
- Expand the university’s knowledge base by incorporating new, structured, and organized data from alumni.
- Streamline processes by using AI to analyze alumni resumes and certifications.
- Maximize employability outcomes through the predictive analysis model enhanced by AI-generated data.
Impact on Marketing: Enhanced Data and Reduced Costs
By applying AI to document parsing, this leading university significantly improves its data volume and quality, enhancing accuracy and performance. These insights create new marketing opportunities and streamline internal operations. Automation reduces the costs associated with manual data entry, making the process more efficient overall.
Time and cost savings resulting from using AI are already apparent and impressive. For parsing certifications, it takes about 2 minutes of a person's time to interpret and catalog a certification. We analyzed about 2,000 certifications in the first round, saving approximately 53 hours (which is 7 working days). For CVs, it took around 5 minutes, and by analyzing about 4,000 CVs, they saved about 40 working days (approximately 330 hours).
