Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Admission Prediction in Undergraduate Applications: an Interpretable Deep Learning Approach

Amisha Priyadarshini, Barbara Martinez-Neda, Sergio Gago-Masague

TL;DR

The paper addresses fairness and scalability in undergraduate admissions by applying two deep neural architectures, FFN and ICNN, to a rich UC Irvine dataset and pairing them with LIME to explain predictions. The authors demonstrate that ICNN achieves competitive accuracy and superior precision, recall, and F1 compared to classical baselines, while a gradient-based feature selection plus LIME provides interpretable insights into which features drive admissions decisions. Positive drivers include GPA and AP performance, whereas negative indicators include the ELC top-percentile measure and EPP participation, supporting a holistic review stance. The work advances transparent, scalable admission decision support and suggests future directions in anomaly detection and model-agnostic explainability to further enhance fairness and accountability in admissions.

Abstract

This article addresses the challenge of validating the admission committee's decisions for undergraduate admissions. In recent years, the traditional review process has struggled to handle the overwhelmingly large amount of applicants' data. Moreover, this traditional assessment often leads to human bias, which might result in discrimination among applicants. Although classical machine learning-based approaches exist that aim to verify the quantitative assessment made by the application reviewers, these methods lack scalability and suffer from performance issues when a large volume of data is in place. In this context, we propose deep learning-based classifiers, namely Feed-Forward and Input Convex neural networks, which overcome the challenges faced by the existing methods. Furthermore, we give additional insights into our model by incorporating an interpretability module, namely LIME. Our training and test datasets comprise applicants' data with a wide range of variables and information. Our models achieve higher accuracy compared to the best-performing traditional machine learning-based approach by a considerable margin of 3.03\%. Additionally, we show the sensitivity of different features and their relative impacts on the overall admission decision using the LIME technique.

Admission Prediction in Undergraduate Applications: an Interpretable Deep Learning Approach

TL;DR

The paper addresses fairness and scalability in undergraduate admissions by applying two deep neural architectures, FFN and ICNN, to a rich UC Irvine dataset and pairing them with LIME to explain predictions. The authors demonstrate that ICNN achieves competitive accuracy and superior precision, recall, and F1 compared to classical baselines, while a gradient-based feature selection plus LIME provides interpretable insights into which features drive admissions decisions. Positive drivers include GPA and AP performance, whereas negative indicators include the ELC top-percentile measure and EPP participation, supporting a holistic review stance. The work advances transparent, scalable admission decision support and suggests future directions in anomaly detection and model-agnostic explainability to further enhance fairness and accountability in admissions.

Abstract

This article addresses the challenge of validating the admission committee's decisions for undergraduate admissions. In recent years, the traditional review process has struggled to handle the overwhelmingly large amount of applicants' data. Moreover, this traditional assessment often leads to human bias, which might result in discrimination among applicants. Although classical machine learning-based approaches exist that aim to verify the quantitative assessment made by the application reviewers, these methods lack scalability and suffer from performance issues when a large volume of data is in place. In this context, we propose deep learning-based classifiers, namely Feed-Forward and Input Convex neural networks, which overcome the challenges faced by the existing methods. Furthermore, we give additional insights into our model by incorporating an interpretability module, namely LIME. Our training and test datasets comprise applicants' data with a wide range of variables and information. Our models achieve higher accuracy compared to the best-performing traditional machine learning-based approach by a considerable margin of 3.03\%. Additionally, we show the sensitivity of different features and their relative impacts on the overall admission decision using the LIME technique.
Paper Structure (15 sections, 5 equations, 3 figures, 2 tables)

This paper contains 15 sections, 5 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Classical Machine Learning in Admission process
  4. Deep Learning in Admission process
  5. Methodology
  6. Data preprocessing
  7. Neural Network Training
  8. Feed-Forward Neural Network
  9. Input Convex Neural Network
  10. Principal Component Analysis
  11. Prediction Interpretability using LIME
  12. Simulation Results
  13. Performance Metrics
  14. Discussion
  15. Conclusion and Future Works

Figures (3)

  • Figure 1: Visual representation of deep neural network architectures, (a) Feed-Forward neural network; (b) Input convex neural network.
  • Figure 2: Confusion matrices for fully-trained deep learning models, (a) Feed-Forward neural network; (b) Input convex neural network.
  • Figure 3: LIME interpretation of the Feed-Forward model predictions. The green and red horizontal bars signify the key features influencing the overall classification in positive and negative ways respectively.

Theorems & Definitions (3)

  • Definition 1: Feed-Forward Predicted Output
  • Definition 2: FICNN Predicted Output
  • Definition 3: Gradient Computation