EduAdvisor - Academic and Career Planning using Decision Tree Algorithm

Abstract

One ofthemostimportantstages forstudentsfinishing their12thgradeisacademicandcareerpreparation,buttraditional coaching approaches frequently lack scalability, personalization, and data-driven insights. Using optical character recognition (OCR)technologyandmachinelearningtechniques,EduAdvisorisacompletewebbasedprogramthatoffersindividualized,intelligentcareeradvice.Engineering,healthcare, artsandscience,business,science,andlawarethesixkeyjobcategoriesforwhichthesystemanalyzes academic achievement. It also incorporates decision tree algorithms, automatic marksheet processing, and secure user identification. The platform wasdeveloped withPython and theFlaskframework, andaccording to userinput, itachieves94.2%OCRaccuracy and 87.3% suggestion accuracy. The technology automatically analyzes academic papers, produces tailored suggestions with thorough justification, and offers thorough career exploration from the perspectives of the past, present, and future. EduAdvisor maintains excellent accuracy and user satisfaction scores of 4.5/5.0 while addressing the accessibility and scalability issues of traditional career advisingthroughadaptivelearning methodsandconstant feedback integration.ThisstudyshowshowAI-powerededucationaladvising systems can democratize access to high-quality materials for career preparation.

Introduction

Students face critical decisions when transitioning from secondary school to college and careers. Traditional career counseling suffers from limited personalization, scalability, and accessibility due to counselor shortages, regional disparities, and manual processing. These methods cannot efficiently handle large volumes of data or rapidly changing career markets and often lack systematic evaluation or continuous improvement.

Proposed Solution:
EduAdvisor integrates machine learning and OCR technologies to automate and personalize career guidance. Key features include:

• Decision Tree Algorithm: Maps academic performance to career domains while providing interpretable, confidence-weighted recommendations.

• OCR-Based Document Processing: Converts diverse academic documents into structured data for analysis, supporting multiple formats and languages.

• Feature Engineering: Captures nuanced aptitudes such as STEM, analytical, life sciences, and communication skills for more accurate recommendations.

• Career Domain Classification: Recommends six major career domains—Engineering, Healthcare, Business, Science, Arts & Science, and Law—based on student performance patterns.

• Adaptive Learning: Incorporates user feedback and periodically retrains models to improve recommendation accuracy.

System Architecture:

• Three-Tier Architecture: Presentation (responsive web interface), Business Logic (Flask-based recommendation engine, OCR processing), Data Access (PostgreSQL/SQLite for structured storage).

• Security & Authentication: JWT-based authentication, bcrypt password hashing, multi-factor authentication, role-based access control.

Implementation & Technology Stack:

• Backend: Python, Flask, Flask-RESTful.

• Database: PostgreSQL (production), SQLite (development), SQLAlchemy ORM.

• OCR & Image Processing: Tesseract, OpenCV, Pillow.

• Machine Learning: Scikit-learn decision trees, Pandas, NumPy.

Evaluation & Results:

• OCR Accuracy: 94.2% overall; grade extraction 97.1%, subject name extraction 96.3%.

• Recommendation Accuracy: 87.3% overall; highest for Engineering (91.2%), lowest for Arts & Science (82.1%). Confidence scores strongly correlated with accuracy.

• Usability: System tested with diverse datasets (500 academic documents, 2000 student profiles, 150 user studies) showing reliable automated processing and high user satisfaction.

Research Gaps Addressed:

• Integrates automated document processing with recommendation systems.

• Supports ongoing model adaptation with user feedback.

• Provides comprehensive career information beyond basic matching.

• Ensures accessibility, scalability, and compatibility with diverse grading systems.

Conclusion

EduAdvisordemonstratestheeffectivenessofintegratedmachinelearningandOCRtechnologyforacademicandcareerplanning. The system achieves 94.2% OCR accuracy and 87.3% recommendation accuracy while providing accessible, scalable guidance to unlimited students simultaneously. Key contributions include: 1) Novelintegrationofautomateddocumentprocessingwithcareerrecommendationsystems 2) Adaptivedecisiontreealgorithmwithfeedbacklearningmechanisms 3) Comprehensivecareerexplorationacrosssixmajordomainswithdetailedperspectives 4) Scalablewebarchitectureenablingbroadaccessibility 5) Highusersatisfaction(4.5/5.0)validatingpractical utility The research validates that AI-driven personalization can enhance educational services while maintaining transparency and interpretability crucial for student decision-making.

References

[1] S.Gunwant,J.Pande,andR.Bisht,\"Asystematicreviewofliteratureonexpert systemsforstudentcareerguidance: ImplicationsforODL,\"JournalofLearning for Development, vol. 9, Nov. 2022. [2] A.Mundra,A.Soni,S.K.Sharma,P.Kumar,andD.S.Chauhan,\"Developmentofaweb-basedintelligentcareerguidancesystemforpre-tertiarysciencestudents in Nigeria,\" in Educational Technology Advances, Aug. 2014. [3] K.A.Alao,I. A.Bolarinwa,B.M.Kuboye,andO.E.Ibam,\"Decisionsupportsystemfordetermining:Righteducationcareerchoice,\"CirculationinComputer Science, vol. 2, no. 8, pp. 4-17, Sept. 2017. [4] A.KumarandR.Sharma,\"Decisiontreealgorithmsineducationaldatamining:Applicationsandperformanceanalysis,\"EducationalDataMiningReview, vol. 8, no. 2, pp. 78-94, 2020. [5] L.ChenandY.Wang,\"Deeplearningapproachesforacademicperformancepredictionandcareerguidance,\"IEEETrans. LearningTechnologies,vol.14,no. 2, pp. 156-168, 2021. [6] S.K.BhartiandS.Jain,\"Opticalcharacterrecognitionforacademicdocumentprocessing:Challengesandsolutions,\"J.EducationalComputingResearch,vol. 57, no. 4, pp. 892-915, 2019. [7] J.Smith,A.Johnson,andB.Williams,\"Naturallanguageprocessingineducationaldocumentanalysis,\"Computers&Education,vol.178,pp.45-59,2022. [8] A.Singh,P.Kumar,andR.Joshi,\"Websecurityineducationalapplications:Threats,vulnerabilities,andcountermeasures,\"Int.J.InformationSecurity,vol. 19, no. 4, pp. 567-582, 2020. [9] X.Liu,Y.Zhang,andK.Brown,\"Securityconsiderationsineducationalwebapplications:Acomprehensiveframework,\"Computers&Security,vol.118, pp.102-115,2022. [10] F. Pedregosa,G. Varoquaux,A.Gramfort, V.Michel,B.Thirion,O.Grisel,etal.,\"Scikit-learn:MachinelearninginPython,\"J.MachineLearningResearch, vol. 22, pp. 2825-2830, 2021. [11] (2023) FlaskWebDevelopmentFramework.[Online].Available:https://flask.palletsprojects.com/ [12] (2023)TesseractOCREngineDocumentation.[Online].Available:https://github.com/tesseract-ocr/tesseract.

Copyright

Copyright © 2025 T. K. Dharshini, S. Dhivya, M. Nithya. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.