In the School of Physics, most courses are assessed with 80% of the final grade based on a May exam and 20% on coursework. Based on a recent survey of all BSc/MPhys students, I analysed students' preferences between exams and coursework using Natural Language Processing.
main.py: Integrate the entire analysis.preprocess.py: Text preprocessing and data extraction.clean_text: Clean text by lowercasing and removing punctuation/stopwords.extract_percentages: Extract explicit and implicit percentage splits (e.g., "40%", "40 exam") and ratio formats (e.g., "40:60").
sentiment_analysis.py: Analyse sentiment in comments related to exams and coursework.get_sentiment: Use TextBlob to classify comments as Positive, Negative, or Neutral based on polarity scores.analyze_sentiments: Separate comments mentioning "exam" and "coursework" and analyse sentiment for each category independently.
topic_analysis.py: Extract reasons for preferences from positive comments.
The results indicate that students prefer coursework over exams. However, the current plot representing the preferred split does not fully reflect the actual feedback, so further investigation is needed.