"Reducing Misaligned University Applications through AI and Machine Learning-Based Recommendation Systems: Insights from a Case Study in Pakistan

Muhammad Ibrahim¹, Muhammad Ahmed², Sara Hania³

¹ Institute of Information Management, University of the Punjab, Lahore, Pakistan
² Institute of Information Management, University of the Punjab, Lahore, Pakistan
³ Department of Computer Science, Alpha College, P.E.C.H.S Block 6, Karachi, Pakistan

Abstract

Every year in Pakistan, thousands of university applicants submit multiple applications without clear guidance, often choosing degrees for which they either lack the required merit or genuine interest. This leads to wasted admission fees, reduced chances of selection, and, in many cases, students ending up in fields they later disengage from—sometimes even abandoning their studies, resulting in wasted years and financial loss.

To address this issue, we developed a basic machine learning–based recommendation model that matches a student’s aggregate score and field of interest with suitable degree programs. The model was built and trained using last-year merit data and categorized interests, with the University of the Punjab serving as a real-world example due to its large variety of programs and high applicant numbers in Lahore as well as other major universities in Pakistan.

While this prototype uses a partial dataset for demonstration purposes, it can be expanded to incorporate full-scale university data. If implemented, the system could be integrated into university application portals, enabling students to receive tailored program suggestions before finalizing their application choices. The web-based interface collects student input (aggregate score and interest category) and returns primary and alternative degree options according to student interest, providing alternatives when merit thresholds are not met. Initial testing on sample data shows promising results. Data suggests that such a system can help students make better-informed decisions, reduce application waste, and improve admission alignment, with potential for adaptation across other universities facing similar challenges.

Keywords — AI in Education; Machine Learning for University Admissions; Degree Recommendation System; Student Career Guidance; Merit-Based Program Selection; Higher Education Technology; Academic Counseling Tools; Pakistan University Admissions; Program Matching Algorithm.

Introduction

University admission represents one of the most critical decisions in a student’s academic and professional life. Selecting an appropriate program is a milestone that significantly shapes an individual’s future. Globally, misaligned program selection has been linked to reduced academic performance, low student satisfaction, and increased dropout rates. In Pakistan, higher education is widely recognized as a key driver of national development, economic growth, and social mobility. However, the system faces persistent challenges that undermine its effectiveness.

A central issue emerges at the initial stage of the educational journey, where a notable disconnect exists between students’ career aspirations and their actual academic choices. Research suggests that the absence of formal career counseling facilities often forces students to rely on family influence or social prestige when making crucial academic decisions, rather than on a realistic assessment of their aptitudes and interests (Shah & Khan, 2018).

This lack of structured guidance during the decision-making period results in several adverse outcomes. Although some universities offer merit calculators or generic career counseling, such tools frequently fail to integrate both merit eligibility and student interest into a unified recommendation. Furthermore, existing approaches rarely provide relevant alternatives when a student’s primary preferences are not feasible. As a result, many students in Pakistan submit multiple applications without a coherent strategy, incurring unnecessary financial costs and increasing the likelihood of enrolling in programs they later disengage from (Aslam & Anwar, 2020). The long-term consequences of these misguided choices are significant, with disengagement and dropout often attributed to a lack of genuine interest (Hussain & Javed, 2019). This contributes to wasted years, financial loss, and inefficiencies within the national education system.

This issue is particularly evident in large, multidisciplinary institutions such as the University of the Punjab. Owing to the wide range of degree programs offered and the absence of structured guidance, students often apply to programs without proper knowledge of requirements or career outcomes. In some cases, students express genuine interest in certain programs but fail to meet the aggregate scores necessary for admission. In others, they apply to programs in which they lack real interest due to a lack of awareness. These patterns lead to repeated misaligned applications, unnecessary admission expenditures, and, in some cases, students abandoning their chosen programs altogether. For instance, in the 2025 undergraduate admission cycle, the University received over 123,000 applications for just approximately 45,000 available seats (Minute Mirror, 2025; University of the Punjab, n.d.). This disparity highlights the scale of misaligned applications and the financial burden imposed on students through repeated form submissions. The University of the Punjab is used here solely as a representative case to illustrate these challenges. The choice of this institution is not intended as a critique but rather as an example to address the broader systemic issues faced by applicants in large universities across the country.

Ultimately, this cycle of inadequate guidance, misinformed academic choices, and subsequent disengagement contributes to broader societal challenges. It exacerbates the existing mismatch between graduate skills and labor market demands, thereby increasing youth unemployment and underemployment (Nadeem & Hassan, 2017). While the literature has addressed dropout rates and skills gaps, there remains a significant research gap in directly examining the university application process and the lack of proper guidance that links to long-term academic and career outcomes.

This paper addresses the research question: How can AI and machine learning effectively predict and guide students toward university programs that align with their capabilities and merit, while also providing relevant alternatives to reduce misaligned applications? To answer this, we propose a machine learning–based program recommendation model that integrates a student’s aggregate score with their declared field of interest. Using last-year merit data and categorized interests, the model generates both primary and alternative program recommendations, with the University of the Punjab serving as a case study. Although the current prototype is based on a partial dataset, it demonstrates the potential for a scalable solution that can be adopted across universities to improve admission alignment and reduce misapplications.

Literature Review

Selecting an appropriate academic program is a crucial step in shaping students’ future careers, yet research has shown that many students worldwide make these decisions without adequate guidance. Arif et al. (2013) found that social pressures, peer influence, and family expectations play a significant role in program selection in South Asia, often at the expense of students’ personal aptitudes and interests. Globally, this trend is evident in studies linking misaligned program choices with low student satisfaction, poor academic performance, and high dropout rates (Nadeem & Hassan, 2017). Collectively, these findings demonstrate that the absence of structured decision-making support mechanisms contributes to academic disengagement, reinforcing the global need for more systematic guidance in university admissions.

In Pakistan, the issue is particularly acute due to the limited availability of formal counseling services at the university level. Hamid Ali (2021) evaluated career counseling facilities in higher education institutions and reported that most universities either lacked such services entirely or offered them in fragmented and poorly resourced forms. Similarly, Yaqoob et al. (2020), in a study on healthcare students, noted widespread dissatisfaction with counseling opportunities, forcing students to rely on informal advice from family or peers. Earlier findings by Ullah et al. (1996) confirm that even when teachers acknowledged the importance of career counseling, institutional neglect prevented its effective implementation. Taken together, these studies highlight the persistent structural gap in Pakistan’s higher education sector, where students are often left without adequate support during one of the most critical academic decisions of their lives. This creates the space for technology-driven interventions that can fill the guidance void.

Recent advances in artificial intelligence (AI) and machine learning (ML) have demonstrated the potential of data-driven approaches to support student decision-making. Kalokhe and Kumbhar (2025) proposed a personalized course recommendation system that matched students’ goals and interests using ML techniques, showing improved satisfaction and alignment with academic paths. Similarly, Esteban et al. (2024) developed a hybrid recommender system with genetic optimization to help university students select electives, demonstrating strong performance in tailoring course suggestions to individual needs. The broader impact of such systems was further highlighted in an MDPI study (2024), which found that AI-driven recommendation tools significantly influenced how students perceived and evaluated course selection decisions. Collectively, this body of research underscores the effectiveness of AI-based recommenders in addressing challenges that traditional counseling methods fail to overcome. However, there remains a gap in applying such technologies at the admissions stage, where students choose their main academic stream.

While AI-driven recommendation systems have shown strong potential internationally, their application in Pakistani higher education remains limited. Akhtar (2020) attempted one of the earliest course recommender implementations at the Virtual University of Pakistan, but its scope was restricted to internal course selection rather than admissions-level guidance—where the real requirement exists. Similarly, Soumalias et al. (2022) demonstrated the role of ML in improving fairness in course allocation, but such approaches have yet to be localized for Pakistani universities. Given the high volume of misaligned applications in institutions such as the University of the Punjab—where 123,000 applicants competed for only 45,000 seats in 2025 (Minute Mirror, 2025; University of the Punjab, 2025)—the absence of structured guidance systems results in wasted resources and financial burdens on students.

This gap highlights the urgent need for an AI-driven admissions recommender that simultaneously considers merit and interest, while also suggesting alternative programs when primary preferences are not feasible. Such a system has the potential to reduce misaligned applications, improve overall academic alignment, and contribute to more efficient higher education processes in Pakistan.

Methodology

The aim of this research is to design a recommendation system that guides students toward university programs aligned with both their academic merit and personal interests. Many students apply to programs either beyond their merit or outside their genuine areas of interest, which often leads to rejection, misaligned career choices, and financial losses from unsuitable enrollments. The proposed methodology reduces this mismatch by applying a structured filtering process that identifies the most suitable programs, suggests alternatives, and highlights potential career prospects.

The overall workflow involves:

Program Data Collection – last year’s merit, categories, alternatives, and career paths.
Student Input Collection – aggregate scores and declared interests.
Filtering Logic Application – step-by-step recommendation of best-matching programs.

Possible Implementation Approaches

Several technical approaches can be adopted depending on data volume and system requirements:

Prototype (Excel + Python scripts): Cost-effective, straightforward, but not scalable.
Database-backed System (MySQL, PostgreSQL, MongoDB): Efficient querying with Python/Node.js backend, requiring server management.
Interactive Frontend (React + REST API in Flask/Django/Node.js): Scalable and mobile-friendly, but higher development effort.
Lightweight Browser-based (JSON/XML + JavaScript): No server dependency, easy to deploy, but limited machine learning capability.
Full-scale Cloud Deployment (Firebase, AWS, Google Cloud): Supports multi-user access and scalability, though introduces costs and internet dependency.

Filtering and Logic

The three-step filtering algorithm is the core contribution of this research. A 3% tolerance margin is included to account for annual merit fluctuations, ensuring students always receive viable recommendations.

1. Primary Filter (Interest + Merit)

The system first identifies programs in the student’s declared interest category where their aggregate score meets or exceeds last year’s merit.

Example:
Student interest = Technology, aggregate = 85%

Computer Science (Merit 86): Eligible (within tolerance)
Software Engineering (Merit 83): Eligible
Information Technology (Merit 80): Eligible
Data Science (Merit 78): Eligible

2. Secondary Filter (Interest but Merit Low)

If no program qualifies under the strict merit rule, the system applies the 3% tolerance margin within the same category.

Example:
Student interest = Medical, aggregate = 70%

MBBS (Merit 90): Not Eligible
Pharm-D (Merit 85): Not Eligible
DPT (Merit 75): Not Eligible
Medical Lab Technology (Merit 71): Eligible (within 1% tolerance)

Recommended: Medical Lab Technology

3. Fall-back Filter (Merit Only, Ignoring Interest)

If no program is feasible within the chosen interest category, the system suggests options purely based on merit eligibility.

Example:
Student interest = Technology, aggregate = 60%

Technology programs (Merit ≥ 75%): Not Eligible
Social Sciences (Merit 60%): Eligible
Arts & Humanities (Merit 55–60%): Eligible

Recommended: Sociology, BBA, Fine Arts, English Literature, Political Science

Outcome of the Filtering Process

After these three phases, the student receives a categorized list of recommendations:

Suitable programs within their declared interest.
Alternative programs within the same domain.
Merit-only fallback programs from other domains.

This structured approach ensures that students are never left without options and supports informed decision-making that balances both academic merit and personal interest.

The following flowchart visually illustrates the three-step filtering process described above, showing how the system recommends suitable, alternative, and fall-back programs based on a student’s aggregate and interest category.

Code Snippet Showing Primary Filtering Logic

Our Approach

In many universities, students often lack proper guidance when selecting degree programs that align with both their academic merit and personal interests. To address this issue, we propose a system that assists students in making informed degree choices through a combination of structured data and machine learning techniques.

For prototyping, we adopted a simple and cost-effective implementation. The dataset (2024) of programs was collected from the University of the Punjab, Lahore, Pakistan, and stored in Excel. Python (Pandas + Flask) was used to process the data and apply the filtering logic. A basic HTML form was developed to collect student inputs such as aggregate and interest category, which were then passed to the Python script. The system applied a three-step filtering process (Primary: Interest + Merit, Secondary: Interest with tolerance, Fall-back: Merit only) and returned the recommended degree programs.

Although this study used a rule-based filtering approach, future implementations may incorporate full-scale machine learning models trained on large admission datasets, enabling more adaptive and intelligent recommendations.

For this study, the dataset from the University of the Punjab (2024) provided program details, categories, and previous-year merit requirements, serving as the basis for implementation and evaluation.

Case Demonstrations of the Filtering System

To illustrate the system’s functionality, we present three scenarios based on varying student aggregates and selected interest categories. Each case demonstrates how the filtering logic (Primary → Secondary → Fall-back) produces tailored recommendations.

Case 1: Primary Filter (Interest + Merit)

Suppose a student has an aggregate of 89% and selects the Technology category. The available programs in this category are:

Data Science (87%)
Information Technology (89%)
Software Engineering (89%)
Computer Science (90%)

Since the student’s merit is 89%, the system recommends Data Science, Information Technology, and Software Engineering, as the student meets or exceeds their merit requirements. Computer Science is not recommended because it requires 90%, which is higher than the student’s aggregate.

This ensures that only programs fully achievable based on the student’s current merit are suggested.

Figure 3.1: Case Demonstration of Primary Filter (Interest + Merit) This figure shows how the system recommends programs like Data Science, Information Technology, and Software Engineering when the student’s aggregate meets or exceeds the required merit in the Technology category.

Case 2: Secondary Filter (Interest with Tolerance)

In another scenario, a student has an aggregate of 70% and selects the Medical category. The available programs are:

Pharm-D (85%)
DPT (78%)
Medical Lab Technology (71%)

Here, the student does not meet the requirements for Pharm-D or DPT. However, the system identifies Medical Lab Technology as a suitable option because the student’s score is only 1% below the required merit, which falls within the 3% tolerance margin.

This demonstrates how the system continues to prioritize the student’s interest area, ensuring that a relevant degree is still recommended even when strict merit thresholds are not met.

Figure 3.2: Case Demonstration of Secondary Filter (Interest with Tolerance) This figure illustrates how the system applies a 3% tolerance margin in the Medical category, recommending Medical Lab Technology when the student’s score is slightly below the merit cut-off

Case 3: Fall-back Filter (Merit only) Finally, consider a student with an aggregate of 60% who selects the Technology category. Since all technology programs require much higher merit, no suitable option is found under the primary or secondary filters. In this situation, the system activates the fall-back filter and recommends degrees from other categories that match the student’s merit level, such as Islamic Studies and Sociology.

Fall-back Filter (Merit Only)

In some situations, a student’s aggregate is too low to qualify for any program within their chosen interest category. In such cases, the system applies the fall-back filter, which ignores interest and instead suggests programs where the student’s merit is sufficient.

Example:
Suppose a student has an aggregate of 60% and selects the Technology category. The available programs in this category all require merit scores of 75% or above, leaving the student ineligible. Instead of returning no results, the system shifts to merit-only filtering and recommends alternative options from other domains, such as Social Sciences (60%) or Arts & Humanities (55–60%).

This ensures that students always receive realistic program suggestions, even if their preferred interest area is not currently achievable.

Test Scenarios and Outcomes

Analysis of Experimental Findings

The experiments revealed several consistent patterns in how the system processed student inputs and generated recommendations:

High Merit and Direct Match
- When students achieved aggregates close to or above competitive cutoff scores, the system successfully recommended programs within their declared interest area.
- This validates the system’s ability to function as a merit-sensitive filter, ensuring students with strong academic records receive relevant program suggestions.
Moderate Merit and Related Alternatives
- For students with aggregates slightly below competitive thresholds, the system identified degree options from closely related fields within the same interest domain.
- This ensured that guidance was not lost, and students still received feasible suggestions that aligned with their broader academic interests.
Low Merit and Backup Pathways
- In cases where students had significantly lower aggregates, the system avoided leaving them without options by recommending fallback programs from other domains.
- While these alternatives were not perfectly aligned with declared interests, they still offered viable academic directions and reduced the likelihood of misaligned or repeated applications.

Key Insights

The experiments demonstrate that integrating merit eligibility and interest categories within a single framework can significantly reduce misaligned applications.
The structured provision of alternative recommendations plays a critical role by ensuring students maintain realistic backup pathways, balancing aspiration with feasibility.
Overall, the findings suggest that the proposed model has the potential to enhance admission alignment, reduce unnecessary application expenditures, and support more informed decision-making among students.
Even at a basic prototype stage, the model shows potential to be deployed as a decision-support system on university portals, reducing wasted application fees and improving student satisfaction in program selection

Analysis

Overview of Results

The proposed University Program Recommendation System demonstrates strong potential in guiding students toward degree programs that align with both their academic merit and declared interests. By employing a structured three-step filtering algorithm with a built-in 3% tolerance margin, the system ensures that students always receive meaningful recommendations rather than being left without viable options.

The experimental findings indicate that the system is effective in addressing several key challenges in the admissions process:

Reduction of Misaligned Applications
The system minimizes the likelihood of students applying to programs where they lack the required merit or genuine interest, thereby reducing rejection rates.
Support for Career Alignment
By recommending programs consistent with student interests and academic performance, the system contributes to better career alignment and long-term engagement with chosen fields.
Financial Impact
The provision of realistic and feasible recommendations lowers the incidence of wasted application fees, a recurring issue in high-volume admission cycles at large universities.

Analysis

Overview of Results

The experimental findings indicate that the system is effective in addressing several key challenges in the admissions process:

Reduction of Misaligned Applications
The system minimizes the likelihood of students applying to programs where they lack the required merit or genuine interest, thereby reducing rejection rates.
Support for Career Alignment
By recommending programs consistent with student interests and academic performance, the system contributes to better career alignment and long-term engagement with chosen fields.
Financial Impact
The provision of realistic and feasible recommendations lowers the incidence of wasted application fees, a recurring issue in high-volume admission cycles at large universities.

Explanation of Outcomes

Primary Filter ensures that students are matched with programs in their declared interest area when their aggregate score meets or slightly exceeds the required merit cut-offs, thereby maximizing relevance and feasibility.
Secondary Filter introduces a 3% tolerance margin, allowing students who fall narrowly below merit thresholds to still receive meaningful program suggestions within their interest domain.
Merit-Only Fall-back Filter guarantees that students are provided with alternative options outside their chosen interest category if no suitable matches are available.

This layered filtering approach ensures wide coverage, reduces the risk of leaving students without guidance, and maintains meaningful recommendation pathways across varying levels of academic performance.

Comparison and Advantages

Efficiency

Unlike traditional counseling mechanisms or earlier recommendation prototypes, the proposed system automates the entire decision-support process and generates recommendations within seconds. Conventional manual counseling requires multiple sessions, consuming significant time and resources and often delaying decision-making.

For example, Booker (2009) [Ref. 2] presented a prototype recommendation system that required heavy manual intervention, leading to longer processing times. Similarly, traditional frameworks reported in MDPI (2022) [Ref. 1] relied on counselor availability, which restricted scalability and slowed down guidance delivery. In contrast, our automated approach is resource-light, rapid, and capable of providing results instantly, making it highly efficient.

Accuracy

The proposed system enhances accuracy by integrating both academic merit and personal interests to deliver precise, personalized recommendations. In contrast, many existing models primarily rely on merit-only filtering, which neglects students’ individual preferences and often results in mismatches.

For instance, ResearchGate (2009) [Ref. 2] describes a prototype system focused solely on academic thresholds, overlooking student interests, thereby risking program misalignment. Likewise, the PMC review (2022) [Ref. 3] stressed that disregarding learner profiles in educational recommender systems significantly reduces the quality of guidance. Our model addresses this gap by ensuring that merit and interest are jointly considered, thus increasing the likelihood of successful academic and career alignment.

Comparison and Advantages

By combining both merit and interest-based filtering, the proposed system ensures that students are guided toward degree programs that best match their profiles, thereby minimizing the risk of later career dissatisfaction. The model holds several advantages over traditional and existing approaches:

1. Simplicity and Accessibility

Unlike traditional counseling methods, where students must physically visit universities or advisors, our system provides fully automated, online recommendations. This eliminates geographical and scheduling barriers, allowing students to access tailored guidance anytime and anywhere.

According to MDPI (2021) [Ref. 4], many institutions continue to rely on in-person counseling, creating bottlenecks for both students and universities. By removing such dependencies, our model delivers instant results without waiting for counselor availability, ensuring seamless accessibility and timely support.

2. Personalization

Most existing recommendation systems are generic and fail to incorporate a student’s unique combination of interests, performance, and long-term goals. For instance, MDPI (2022) [Ref. 1] highlights that many current recommenders remain largely rule-based, offering limited personalization.

Our system addresses this limitation through a three-step personalized filtering mechanism (Primary → Secondary → Fall-back), which tailors results for each individual student. This personalization not only increases satisfaction but also improves the likelihood of sustained academic engagement and long-term career success.

3. Cost-Effectiveness

Traditional counseling services and semi-automated recommenders typically involve significant financial investments, including counselor fees, institutional infrastructure, and high-maintenance computing systems. For example, ResearchGate (2023) [Ref. 5] shows that many user-based recommenders require expensive storage and computational resources, particularly when operating on large datasets.

In contrast, our system is lightweight, operating with minimal computational cost while remaining scalable without heavy investments. This makes it a cost-effective solution for both students and institutions. Beyond institutional savings, the system also reduces financial burdens for families: in the 2025 undergraduate admission cycle at the University of the Punjab, over 123,000 applications were submitted for approximately 45,000 available seats, reflecting significant waste in repeated form submissions and application fees. (Minute Mirror, 2025; University of the Punjab, n.d.). With each application costing around PKR 500, this disparity translates into nearly PKR 39 million in wasted application fees for students whose forms did not lead to admission. Such losses, which also occur across other major universities in Pakistan, highlight the urgency of structured guidance mechanisms that can reduce misaligned applications and the associated financial waste. By helping students target programs that match both merit and interest, our system not only improves academic alignment but also prevents large-scale economic losses borne by families

By streamlining program selection, the system helps mitigate these recurring costs while improving admission efficiency.

Conclusion and Future Work

This study addressed the persistent issue of misaligned university applications in Pakistan by proposing a machine learning–based recommendation model that aligns students’ merit and interests with suitable degree programs. Using the University of the Punjab as a representative case, the research demonstrated how merit-aware and interest-based guidance can reduce wasted applications, financial burdens, and subsequent academic disengagement.

The results indicate that such a system can provide meaningful decision support to students, enabling more informed program choices and reducing inefficiencies in the admissions process. Even at the prototype stage, the model illustrates the potential of AI-driven tools to reshape higher education admissions by combining accessibility, personalization, and cost-effectiveness.

Nonetheless, the current implementation is limited by its reliance on a partial dataset and a relatively simple algorithm. Future research should broaden the dataset across multiple universities, incorporate advanced machine learning methods for deeper personalization, and integrate additional dimensions such as career trajectories, labor market trends, and socio-economic constraints. By advancing in these directions, universities can move toward a more structured and student-centered admissions framework that not only improves academic alignment but also contributes to long-term career satisfaction and national human capital development.

References
Akhtar, N. (2020). Course recommender system for Virtual University of Pakistan. International
Journal of Advanced Computer Science and Applications, 11(5), 34–40.
Ali, H. (2021). Evaluation of career counseling services in higher education institutions of Pakistan.
Journal of Education and Educational Development, 8(2), 290–307.
Arif, I., Chaudhry, N., & Raza, H. (2013). Factors influencing students’ choice in higher education:
A South Asian perspective. International Journal of Educational Research, 58, 46–55.
https://doi.org/10.1016/j.ijer.2012.12.002
Esteban, G., Domínguez, C., & De La Fuente, J. (2024). A hybrid recommender system with genetic
optimization for course selection in higher education. Journal of Educational Technology & Society,
27(1), 1–15.
Hamid Ali. (2021). Career counseling at university level: Challenges and opportunities in Pakistan.
Asian Journal of Education and Training, 7(3), 187–194. https://doi.org/10.20448/journal.522.2021.73.187.194
Hassan, A., Iqbal, S., Mazhar, S., & Dogar, S. F. (2025). Evaluating the effectiveness of career counseling services in Pakistani universities. International Journal of Education Research and Development, 7(2), 15–28.
Hussain, S., Hussan, K. H. ul, & Ahmed, M. (2023). Exploring the need of career counselling for
choosing the career field at school level in Pakistan: Parents’ and teachers’ perceptions. Global Educational Studies Review, 8(1), 143–152.
Kalokhe, A., & Kumbhar, R. (2025). Personalized course recommendation system using machine
learning. International Journal of Computer Applications, 183(23), 12–18.
Keshf, Z. (2022). “It is a very difficult process”: Career service providers’ perspectives on career
counseling needs. Frontiers in Education, 7, 856632. https://doi.org/10.3389/feduc.2022.856632
Keshf, Z., & Khanum, S. (2022). Career guidance and counseling services in Pakistan from the perspective of students and career service providers. Pakistan Journal of Psychological Research, 37(1),
39–66. https://doi.org/10.33824/PJPR.2022.37.1.39
Minute Mirror. (2025, January 6). PU says yes to bright future with 123K undergraduate admissions.
Minute Mirror. https://minutemirror.com.pk/pu-says-yes-to-bright-future-with-123k-undergraduateadmissions-423275/
Mdpi. (2021). Institutions and the availability of academic programs. Social Sciences, 5(2), 33.
https://www.mdpi.com/2504-4990/5/2/33
Mdpi. (2022). Approximately 30% of first-year students face challenges. Applied Sciences,
12(24), 12525. https://www.mdpi.com/2076-3417/12/24/12525
Nadeem, A., & Hassan, S. (2017). Impact of mismatched academic choices on student satisfaction and dropout rates. Higher Education Studies, 7(2), 120–129.
https://doi.org/10.5539/hes.v7n2p120
PMC. (2022). Educational recommender systems in personalized learning. Journal of Personalized Medicine, 12(8), 1295. https://doi.org/10.3390/jpm12081295
ResearchGate. (2013). A student program recommendation system prototype. https://www.researchgate.net/publication/255641503_A_STUDENT_PROGRAM_RECOMMENDATION_SYSTEM_PROTOTYPE
ResearchGate. (2023). An approach to implement user-based recommendation systems with
small-sized data sets. https://www.researchgate.net/publication/376892911_An_approach_to_implement_user-based_recommendation_systems_with_small-sized_data_sets
Soumalias, A., Karampelas, P., & Tsiatsos, T. (2022). Machine learning approaches for fair
course allocation in higher education. Education and Information Technologies, 27, 6523–6540.
https://doi.org/10.1007/s10639-021-10826-2
Ullah, H., Akbar, R., & Farooq, S. (1996). Teachers’ perspectives on career counseling in Pakistan. Pakistan Journal of Education, 13(1), 23–37.
University of the Punjab. (n.d.). How many on-campus full-time students are enrolled in the
university’s six campuses? University of the Punjab. https://pu.edu.pk/faq/ans/48
University of the Punjab. (n.d.). PU at a glance. University of the Punjab.
https://pu.edu.pk/page/puglance/
University of the Punjab. (n.d.). Undergraduate programs merit of last year 2024. Directorate
of Students Affairs. https://pu.edu.pk/downloads/Merit-of-Last-Year-2024.pdf
Yaqoob, S., Saeed, A., & Qureshi, M. (2020). Perceptions of healthcare students regarding career
counseling in Pakistan. Pakistan Journal of Medical Sciences, 36(2), 150–156.
https://doi.org/10.12669/pjms.36.2.1181

Search This Blog

STEM-POWERED Pakistan