Career Confusion Prediction System for Students

Career Confusion Prediction System for Students

Sujeetha D,

Assistant Professor,

Department of Computer Science and Engineering Nehru Institute of Engineering and Technology Coimbatore -641105

Sreerooga M, Kirthick Rosan B, Meera Rose Michel M, Anagha K V

Department of Computer Science and Engineering, Nehru Institute of Engineering and Technology Coimbatore -641105

Abstract – Career decision-making is a critical phase in a students academic journey, often accompanied by uncertainty and confusion due to multiple available options and unclear self-assessment. Traditional career guidance systems primarily focus on aptitude testing or preference-based recommendations, without evaluating the stability of a students decision-making process. This paper proposes a Career Confusion Prediction and Roadmap Recommendation System that first identifies the level of career confusion and then provides a structured career pathway. The system analyzes behavioral indicators such as response hesitation, answer variability, interest switching, and expectation mismatch to compute a confusion score. Based on this score, students are classified into Low, Medium, or High confusion levels using a rule-based decision framework. For students with stable decision patterns, the system suggests a final career path and generates a detailed study roadmap outlining educational steps, skill development stages, and progression milestones.

The proposed approach serves as an integrated decision-support framework that not only predicts confusion but also guides students toward a clear and actionable career direction, promoting informed and confident career choices.

Keywords: Career Confusion Prediction, Behavioral Analysis, Rule-Based AI, Career Recommendation, Roadmap Generation, Decision Support System

1. INTRODUCTION

   Choosing an appropriate career path is one of the most significant decisions in a students academic and professional life. With the rapid expansion of educational opportunities and emerging career domains, students are exposed to a wide range of options. While this diversity provides flexibility, it also creates uncertainty and confusion. Many students struggle to make confident decisions due to conflicting interests, external influences, unclear self-assessment, and unrealistic expectations regarding salary or effort. Traditional career guidance approaches primarily rely on aptitude tests, academic performance, or direct preference-based questionnaires. Although these methods provide recommendations, they often fail to assess the stability and clarity of a students decision-making process. A student may select a career option without being confident or consistent in their responses, leading to future dissatisfaction or career shifts. Therefore, identifying the level of career confusion before suggesting a final path becomes essential. This research proposes a Career Confusion Prediction and Roadmap Recommendation System that integrates behavioral analysis with rule-based decision logic. The system evaluates factors such as response hesitation, answer variability, interest switching patterns, and expectation mismatch to calculate a structured confusion score. Based on the predicted confusion level, the system either stabilizes the students decision-making process or recommends a final career path along with a structured study roadmap. By combining confusion detection, career recommendation, and roadmap generation into a single framework, the proposed system aims to support students in making informed, confident, and well-planned career decisions.

2. PROBLEM STATEMENT

   Career selection has become increasingly complex due to the rapid growth of educational streams and professional opportunities. Students are required to make important career decisions at an early stage, often without complete self-awareness, clarity of interests, or understanding of long-term commitments. While many career guidance systems provide recommendations based on aptitude tests, academic performance, or stated preferences, they do not evaluate the underlying confusion or instability in a students decision-making process. In many cases, students change their choices frequently, hesitate while responding to career-related questions, or express expectations that do not align with their willingness to invest time and effort. Such behavioral inconsistencies

   indicate confusion, which traditional systems fail to measure. As a result, students may receive career suggestions without first addressing their uncertainty, leading to dissatisfaction, poor academic performance, or later career shifts.

   Therefore, there is a need for a system that not only recommends a suitable career but also identifies and quantifies the level of career confusion before providing guidance. Additionally, students often lack a clear roadmap outlining the educational and skill development steps required to achieve their chosen career. Addressing both confusion detection and structured pathway guidance forms the core problem this research aims to solve.

3. LITERATURE SURVEY

   Recent studies show significant developments in web-based career guidance and intelligent recommendation systems:

   • Web-based career guidance platforms primarily utilize aptitude tests and academic performance metrics to suggest suitable career paths. These systems apply clustering and classification techniques to categorize students into predefined career domains, demonstrating structured decision support capabilities [1].
   • Data mining algorithms such as K-Means clustering and decision tree classifiers have been widely adopted in educational recommendation systems to group students based on academic scores and interest patterns. These techniques improve categorization efficiency but rely heavily on explicit user inputs rather than behavioral interaction data [2].
   • Rule-based expert systems have also been implemented to simulate counselor-driven recommendations. Such systems provide deterministic outputs based on predefined logic rules and domain mappings. While effective in structured scenarios, they do not evaluate decision instability or uncertainty in user responses [3].
   • Recent research in behavioral analytics highlights the importance of analyzing interaction patterns, response hesitation, and preference variability to understand cognitive uncertainty in decision-making systems [4]. These approaches suggest that behavioral indicators can offer deeper insight into user confusion beyond traditional test-based evaluation.
   • Additionally, personalized recommendation frameworks increasingly incorporate structured guidance models to provide pathway-oriented suggestions. However, most existing systems limit their functionality to career recommendation and do not generate detailed educational roadmaps outlining progressive milestones [5].
4. EXISTING SYSTEM

   The existing system described in the uploaded paper titled Web Based Career Guidance System published in International Research Journal of Engineering and Technology (IRJET, 2021) primarily focuses on aptitude-test-based career recommendation for students after 10th and 12th grade

   1. Traditional Existing System
      • One-to-one manual counseling
      • Pen-and-paper aptitude tests
      • Guidance based only on academic marks
      • High dependency on counselors These systems had limitations such as:
      • Low accessibility
      • Availability constraints
      • Lack of scalability
      • Limited personalization
   2. Web-Based Existing System (IRJET Model)
      • Student Registration & Login
      • Selection of Grade (10th / 12th)
      • Academic Marks Entry
      • Aptitude Test (Quantitative, Logical, Verbal)
      • Score-Based Career Recommendation
      • Data Mining using K-Means Clustering
      • College List Recommendation
      • Report Generation
      • Backend Database Storage (MySQL via XAMPP)
   3. Limitations of the Existing System
      • Recommendation depends mainly on test scores.
      • No behavioral instability analysis is performed.
      • No confusion quantification mechanism.
      • No measurement of decision hesitation or domain switching.
      • No salary-expectation analysis.
      • No structured roadmap generation for career progression.
      • Requires database storage and backend infrastructure.
      • Does not classify students based on decision stability.
5. PROPOSED SYSTEM OVERVIEW
   1. 1. System Goals

         The primary goal of the proposed Career Confusion Prediction and Roadmap Recommendation System is to assist students in making stable, informed, and structured career decisions. The system is designed with the following objectives:

         • To identify and quantify the level of career confusion in students using behavioral analysis.
         • To evaluate response instability, hesitation patterns, interest switching, and expectation mismatch.
         • To classify students into Low, Medium, or High confusion levels using a rule-based decision framework.
         • To recommend a final career path based on dominant interest patterns and decision stability.
         • To generate a structured educational roadmap outlining step-by-step progression toward the recommended career.
         • To operate using session-based processing without long-term data storage, ensuring privacy preservation.
      2. System Architecture

         Figure 5.2.1: Architecture Diagram

         The system consists of the following core components:

         • User Interaction Module: This module collects student inputs through structured questionnaires and preference-based selections. It records:
         • Career interests
         • Salary expectations
         • Willingness to pursue higher education
         • Response patterns and hesitation
         • All inputs are processed within a session-based environment.
         • Behavioral Analysis Engine: This component analyzes interaction behavior to detect:
         • Answer instability
         • Interest switching frequency
         • Response hesitation
         • Expectationeffort mismatch
         • These behavioral indicators help in identifying cognitive uncertainty.
         • Confusion Scoring and Classification Module: A rule-based decision mechanism computes a numerical confusion score based on behavioral parameters. Students are classified into:
         • Low Confusion
         • Medium Confusion
         • High Confusion
         • Career Recommendation Module: For students with stable decision patterns, the system identifies the dominant career domain and generates a final career suggestion using predefined logical mapping rules.
         • Roadmap Generation Module: This module provides a structured study and skill development roadmap including:
         • Required educational qualifications
         • Certification paths
         • Skill enhancement stages
         • Internship or training suggestions
         • Career progression milestones

   Figure 5.2.2: Career Prediction

6. MATHEMATICAL MODEL FOR CONFUSION SCORE

   The proposed system quantifies career confusion using a structured behavioral scoring mechanism. Instead of relying solely on explicit user selections, the model evaluates implicit behavioral indicators observed during interaction with the system. The confusion score is computed by aggregating multiple normalized behavioral parameters that reflect instability in decision-making.

   1. 1. Four Primary Dimensions:
         1. Answer Instability

            This parameter measures the consistency of responses across related questions. If a student selects conflicting domains in similar question contexts, instability is recorded. Higher inconsistency contributes to a higher confusion score.

         2. Interest Switching Frequency

            This measures how frequently a student changes preferred career domains within a session. Frequent switching indicates uncertainty and increases the confusion level.

         3. Response Hesitation

            This parameter evaluates the time taken to respond to key decision-making questions. Excessive hesitation suggests lack of clarity or confidence, thereby increasing confusion weight.

         4. ExpectationEffort Mismatch

         This evaluates the alignment between a students salary expectations and their willingness to pursue required educational effort. A significant mismatch indicates unrealistic planning, contributing to confusion.

      2. Confusion Score Computation:

         Each parameter is normalized to a standard scale to ensure uniform contribution. The system assigns predefined weights to each parameter based on their relative importance in reflecting decision instability. These weighted values are aggregated to generate a final confusion score.

         The resulting score represents the overall decision stability of the student during the session.

      3. Confusion Level Classification

         The aggregated confusion score is categorized into three levels:

         • Low Confusion Indicates stable and consistent decision patterns.
         • Medium Confusion Indicates moderate uncertainty requiring guided clarification.
         • High Confusion Indicates significant instability; recommendation is deferred until re-evaluation.
      4. Features
   • Multi-Dimensional Evaluation
   • Behavioral-Based Measurement
   • Weighted Parameter Aggregation
   • Normalized Scoring Mechanism

      

   • Real-Time Computation
7. METHODOLOGY
   1. 1. Research Design

         The system follows an Applied and Exploratory Research Design. It focuses on:

         • Behavioral pattern observation
         • Decision instability measurement
         • Confusion quantification
         • Rule-based career recommendation

           The methodology does not depend solely on academic marks but evaluates psychological uncertainty through structured interaction.

      2. System Workflow

         Phase 1: Data Collection (Session-Based Input)

         • Behavioral questionnaire
         • Domain preference selection
         • Salary expectation input
         • Study effort willingness selection

           No long-term database storage is used. All computations occur within the session.

           Phase 2: Behavioral Parameter Extraction

           The system extracts measurable indicators:

         • Answer inconsistency
         • Domain switching frequency
         • Response hesitation
         • Expectationeffort mismatch

           Phase 3: Confusion Score Computation

           A weighted aggregation mechanism is applied to compute a final confusion score. The score represents the degree of decision instability.

           Classification is performed into:

         • Low Confusion
         • Medium Confusion
         • High Confusion

           Phase 4: Career Recommendation Engine

           If the confusion level is low or moderate, the system:

         • Identifies the dominant career domain
         • Maps it to a predefined rule-based career option

           No machine learning model is required. The system operates using deterministic logical mapping.

           Phase 5: Roadmap Generation

           After career selection, a structured roadmap is generated including:

         • Required educational path
         • Skill development stages
         • Certification recommendations
         • Internship guidance
         • Career growth milestones

   This ensures actionable guidance rather than sample recommendation.

8. IMPLEMENTATION PLAN
   1. 1. Technology Stack
         • Frontend: HTML5, CSS3, JavaScript
         • Logic Processing: JavaScript (Client-side)
         • Storage: Session-based (No Database)
      2. Module Development Plan
         • Module 1: User Interface Development
         • Module 2: Behavioral Analysis Module
         • Module 3: Confusion Scoring Engine
         • Module 4: Recommendation Engine
         • Module 5: Roadmap Generator
      3. Testing Plan
         • Functional Testing: Validate each module independently
         • Behavioral Accuracy Testing: Simulate stable and unstable user responses
         • Performance Testing: Ensure real-time score computation
         • Usability Testing: Validate user experience
      4. Deployment Plan
         • Host as a standalone web application
         • No backend server dependency
         • Runs on modern browsers
      5. Methodology Advantage
         • Psychological confusion quantification
         • Behavioral-driven analysis
         • Decision stability validation
         • Structured roadmap generation
9. RESULTS & DISCUSSION
   1. 1. Experimental Setup

         The proposed system was evaluated using simulated user sessions representing different behavioral patterns. The evaluation focused on measuring the systems ability to:

         • Detect decision instability
         • Compute confusion levels accurately
         • Generate appropriate career recommendations
         • Provide structured career roadmaps

           Test cases were designed to simulate three categories of users:

         • Stable decision-makers
         • Moderately uncertain students
         • Highly confused students
      2. Confusion Score Evaluation

         The confusion scoring mechanism successfully differentiated students based on behavioral consistency. Observed Results:

         • Students with consistent domain preference and realistic expectations were classified as Low Confusion.
         • Students showing minor domain switching or moderate hesitation were classified as Medium Confusion.
         • Students with frequent switching, inconsistent answers, and unrealistic expectations were classified as High Confusion.

           The classification aligned logically with the simulated behavioral patterns, validating the theoretical model.

      3. Career Recommendation Accuracy

         For Low and Medium confusion levels:

         • The system successfully identified dominant domains.
         • Rule-based mapping generated consistent career recommendations.
         • Recommendations were explainable and transparent.

           For High confusion level:

         • The system deferred final career recommendation.
         • Users were advised to re-evaluate preferences before final decision.
      4. Roadmap Generation Analysis

         For recommended careers, the roadmap module generated:

         • Educational qualification path
         • Skill development milestones
         • Certification guidance
         • Internship suggestions
         • Career growth progression

           This provided actionable guidance instead of only naming a career stream.

      5. Comparative Discussion
         • Existing systems recommend careers directly based on test performance.
         • The proposed system validates decision stability prior to recommendation.
         • Confusion quantification adds a psychological evaluation layer.
         • No database dependency increases portability and privacy.
      6. System Effectiveness

         The proposed approach demonstrated:

         • Effective behavioral classification
         • Transparent decision-making process
         • Real-time computation capability
         • Structured and explainable output

           The confusion score model proved effective in distinguishing stable decisions from uncertain ones, thereby improving the reliability of career recommendatins.

10. LIMITATIONS

1. 1. • Self-Reported Bias: The system depends on user-provided responses; inaccurate or socially desirable answers may affect confusion score reliability.
      • Behavioral Simplification: Complex psychological factors influencing career decisions are approximated using limited behavioral indicators.
      • Lack of Longitudinal Tracking: Session-based operation without database storage prevents long-term behavioral monitoring and trend analysis.
      • Rule-Based Rigidity: The deterministic rule-based recommendation engine cannot dynamically adapt to new career trends or evolving job markets.
      • Domain Coverage Constraints: The roadmap generator supports predefined career paths; emerging or interdisciplinary careers require manual updates.
      • Scalability of Validation: Large-scale real-world validation across diverse student populations is yet to be conducted.
      • Cultural and Socioeconomic Factors: External influences such as financial background, family pressure, and geographic limitations are not incorporated into the decision model.

        12. FUTURE SCOPE

      • Machine Learning Integration: Incorporating adaptive learning models to dynamically refine confusion scoring and career recommendation accuracy based on real-world user data.
      • Personality Profiling Integration: Adding psychometric assessments (e.g., interest and personality analysis) to enhance behavioral depth and recommendation reliability.
      • Longitudinal Tracking Module: Introducing optional secure database storage to monitor student decision stability over time and detect behavioral evolution.
      • Adaptive Weight Optimization: Implementing automated parameter weight tuning to improve confusion score precision across diverse student populations.
      • Expanded Career Domain Coverage: Including emerging fields such as Artificial Intelligence, Data Science, Renewable Energy, and interdisciplinary careers.
      • Mobile Application Deployment: Developing cross-platform mobile applications for wider accessibility and real-time interaction tracking.

13. CONCLUSION

This study presents a novel behavioral-driven framework for career confusion prediction and structured roadmap generation. Unlike traditional career guidance systems that rely primarily on aptitude scores and academic performance, the proposed model introduces a quantifiable confusion scoring mechanism to evaluate decision stability prior to recommendation. By incorporating behavioral indicators such as response consistency, domain switching patterns, hesitation behavior, and expectationeffort alignment, the system transforms qualitative psychological uncertainty into a measurable computational metric. The rule-based recommendation engine ensures transparency and interpretability, while the roadmap generation module provides actionable academic and skill development guidance beyond mere career identification. The session-based architecture further enhances privacy and deployability without dependency on persistent databases. Experimental validation through controlled simulations demonstrates the models capability to distinguish varying levels of decision instability and provide appropriate guidance accordingly. Overall, the proposed system contributes a structured, interpretable, and research-oriented approach to intelligent career guidance, bridging behavioral analytics with decision-support system design.

ACKNOWLEDGEMENTS

The authors express their sincere gratitude to the project supervisor and faculty members for their guidance, support, and constructive feedback throughout the development of this research work. Their valuable insights significantly contributed to the successful completion of this study.

REFERENCES

1. R. Ghatugade, G. Chormunge, V. Kumar, and R. Nikhare, Web Based Career Guidance System, International Research Journal of Engineering and Technology (IRJET), vol. 8, no. 3, pp. 24902494, Mar. 2021.
2. R. Rangnekar, K. P. Suratwala, S. Krishna, and S. Dhage, Career Prediction Model Using Data Mining and Linear Classification, in Proc. IEEE International Conference, 2018.
3. L. Prasanna and D. Haritha, Smart Career Guidance and Recommendation System, International Journal of Engineering Development and Research (IJEDR), 2019.
4. K. F. Mohammed, S. Gawade, and V. Nimkar, Online Career Counseling System using Adaptive e-Learning, in Proc. International Conference on Recent Innovations in Signal Processing and Embedded Systems (RISE), 2017.
5. S. Kolekar, M. Surve, P. Redkar, and S. Bojewar, A Review: E-Counseling, International Journal of Scientific Research in Computer Science, Engineering and Information Technology, vol. 3, no. 3, 2018.
6. S. K. Yadav, S. J. Singh, A. M. Bora, and S. N. Thakur, Educational Recommendation and Tracking System, International Journal of Scientific & Engineering Research, vol. 7, no. 11, pp. 14431447, 2016. [7] S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 3rd ed. Pearson Education, 2010. [8]

J. Han, M. Kamber, and J. Pei, Data Mining: Concepts and Techniques, 3rd ed. Morgan Kaufmann, 2011.

1. N. S. Abdulwahed, M. Nagy, and H. H. Youssef, Intelligent Agent Based Career Guidance System Using Data Mining and Expert Rules, International Journal of Computer Applications, vol. 150, no. 1, pp. 2127, Oct. 2016.
2. P. S. Patil and S. R. Biradar, Web Based Student Counseling System Using Rule Based Expert System, International Journal of Computer Science and Mobile Computing, vol. 7, no. 2, pp. 102107, Feb. 2018.
3. A. Sharma and N. Kaur, A Hybrid Career Guidance System Using Machine Learning and Decision Support Tools, International Journal of Innovative Technology and Exploring Engineering, vol. 8, no. 10, pp. 25112515, Aug. 2019.
4. R. A. R. Oliveira and C. R. S. de Souza, Behavioral Analytics for Student Success Prediction Using Learning Patterns, Journal of Educational Data Mining, vol. 12, no. 3, pp. 123, 2020.
5. S. K. Mohapatra and B. P. Sahu, Personalized Career Recommendation System Based on User Profile and Job Market Trends, Procedia Computer Science, vol. 167, pp. 18111820, 2020.
6. M. A. Jabbar and S. A. Jawad, Rule-Based Expert System for Enhancing Student Career Guidance and Future Scope Prediction, International Journal of Computer Science and Information Security, vol. 18, no. 4, pp. 5460, 2020.
7. V. R. Shinde and S. S. Ganorkar, AI-Driven Adaptive Recommendation System for Student Skills Development, International Journal of Artificial Intelligence and Applications (IJAIA), vol. 11, no. 2, pp. 6781, Mar. 2021.
8. T. Nguyen and J. Qi, A Comparative Study of Machine Learning Techniques for Career Recommendation Based on Student Data, IEEE Access, vol. 9, pp. 102345102357, 2021.
9. A. Singh and M. Gupta, Deep Learning Approaches for Student Dropout Risk Prediction in Educational Systems, IEEE Transactions on Learning Technologies, vol. 14, no. 1, pp. 7485, Jan.Mar. 2022.
10. M. A. Farooq, A. Arshad, and S. Khan, Decision Support System for Career Selection Based on Sentiment Anlysis of Student Feedback, International Journal of Advanced Computer Science and Applications (IJACSA), vol. 13, no. 6, pp. 390398, 2022.
11. J. C. Chen, C. Y. Chang, and H. Y. Lee, Session-Based Behavior Analytics for Intelligent Tutoring Systems, Educational Technology & Society, vol. 25, no. 1, pp. 89101, 2022.
12. Q. Liu, X. Zhang, and R. H. Deng, Explainable AI for Student Decision Support: A Survey, ACM Computing Surveys, vol. 55, no. 3, pp. 61:161:34, May 2023