Projects
Data-Driven Skilling for Odisha's Workforce
Odisha’s skilling systems generate substantial administrative data across secondary schooling, vocational training, and technical education. Historically, these datasets have operated in silos — making it impossible to track student trajectories across institutions, identify where young people exit the system, or assess whether skilling investments are translating into outcomes.
DPIC, in partnership with the Skill Development & Technical Education and Higher Education Departments, is building Odisha’s first longitudinal skilling data architecture — integrating population-scale administrative datasets to enable evidence-based human capital planning.
Curating Population-Scale Administrative Data
At the core of this initiative is the integration of nearly 500,000 student-level records annually since 2018, spanning more than 25 GB of structured data. DPIC has built privacy-preserving pipelines to consolidate and link records from:
- Applications and admissions from the Student Academic Management System (SAMS).
- Vocational training performance and dropout records from SCTEVT.
- Class 10 (BSE) and Class 12 (CHSE) examination data.
These datasets are being cleaned, standardized, and longitudinally linked to create a unified integrated skilling database — enabling individual-level tracking across the full skilling pipeline.
From Fragmented Records to Skilling Analytics
The integrated database allows government to generate first-time analytical insights previously unattainable from siloed systems:
- Mapping student transition pathways — who continues after secondary school, who pivots to vocational training, and who exits the system altogether.
- Identifying dropout and attrition hotspots — pinpointing where and when students are most likely to leave the skilling pipeline.
- Detecting demand-supply imbalances in skilling — where student applications and enrolment patterns diverge from available training capacity across institutes and trades.
- Diagnosing gender and geographic disparities — in access, retention, and completion across districts and population groups.
- Evaluating programmes on measurable outcomes — shifting assessment from seat utilisation to evidence of completion and placement.
This shifts skilling policy from input-driven seat expansion to evidence-based capacity alignment — where decisions are grounded in what the data reveals about transitions, gaps, and outcomes.
Understanding the Why: Field Research
Administrative data reveals where students drop out or fail to place — but not why. To understand the motivations and constraints shaping student decisions, DPIC in collaboration with University of Chicago researchers is conducting structured field surveys and qualitative studies alongside the administrative data work.
This research examines the drivers of course and trade selection, perceptions of vocational versus academic pathways, financial and social constraints on participation, and barriers to programme completion and placement. By linking survey findings with longitudinal administrative records, the project deepens causal understanding and strengthens the design of targeted policy interventions.
From Skilling Data to Decision Intelligence
The current focus is on building clean, integrated longitudinal datasets and analytical dashboards that provide actionable visibility into the skilling pipeline. As the database matures, it will support the use of AI tools to generate rapid insights — on dropout patterns, trade demand, enrollment trends, and program performance – moving toward real-time decision intelligence for skilling policy.
Alongside the data work, findings from the field research will inform the design of targeted interventions addressing the specific barriers students face in accessing, completing, and benefiting from skilling programs.