Sentiment Analysis of Student Feedback Using NLP Techniques

Sentiment Analysis of Student Feedback Using NLP Techniques

Computer Science

Eshan Sharif

2024

Dr Sunar Ayse

This project focuses on applying sentiment analysis to student feed- back from teaching lessons, employing natural language processing (NLP) techniques such as VADER and TextBlob to classify senti- ments as positive or negative. The dataset, sourced from a univer- sity in the Netherlands, exhibited a significant class imbalance, with a majority of positive sentiments. To address this issue, synthetic data generation via OpenAI’s batch request API was implemented to oversample the minority class (negative sentiments), ensuring a more balanced dataset for training. Post-balancing, machine learning models, particularly Random Forest, were employed to classify the data. The model achieved strong performance, with a balanced accuracy of 99 This report provides a detailed overview of the methodology used, including data preprocessing, model training, and evaluation. It also explores the limitations of the approach and proposes poten- tial directions for future research, including the use of alternative classifiers and more sophisticated text representation techniques to further enhance the accuracy and robustness of sentiment analysis in educational feedback contexts.

Sentiment Analysis
Natural Language Processing
Student Feedback
Educational Feedback
OpenAI API
Synthetic Data Generation
Machine Learning
Class Imbalance
Random Forest
VADER
TextBlob
Technical University of Denmark
URSS Project