mlmi16 — Advanced HCI (CW3)

Coursework materials for the personalisation & inspiration user study: a between-subjects comparison of personalised vs generic LLM suggestions across travel and cooking scenarios, with survey and behavioural outcomes analysed in Python.

Project structure

mlmi16/
├── README.md
├── experiment/
│   └── index.html                    # Study web app (HTML/CSS/JS)
├── data/                             # Intentionally not public: participant dataset removed from this repository
└── analyses/
    ├── analyse.py                    # Cleaning, composites, assumption checks, hypothesis tests, plots
    └── plots/                        # Generated — result figures (PDF)
        ├── fig1_dvs_by_condition_scenario.pdf
        └── fig2_manipulation_check.pdf

Data availability

data/user_study_MLMI16.csv is intentionally not included in this public repository because participant-level study data is not publicly shareable.

Running the analysis

From the repository root (this directory):

python3 analyses/analyse.py

Dependencies: pandas, numpy, scipy, matplotlib. For full output (logistic regression, ANCOVA, mixed ANOVA), also install statsmodels and pingouin:

pip install pandas numpy scipy matplotlib statsmodels pingouin

The analysis script expects data/user_study_MLMI16.csv to be present.