Road Casualty Severity Analysis

I have the 2024 road-casualty dataset already sitting in Google Colab with the raw CSV loaded and some basic data-quality checks (shape, duplicated rows, obvious null counts) completed. From here, I need to take the work through to a submission-ready assignment that meets the brief and marking rubric supplied by my course.

Scope of work
• Polish the preprocessing pipeline: tackle any remaining missing-value issues, tidy categorical variables, and prepare features so they are ready for modeling both in Python (Pandas, scikit-learn) and, if easier for you, in WEKA (ARFF export).
• Carry out three clearly defined lines of analysis that shed light on casualty severity. You are free to choose the angles provided they are relevant, insightful, and defensible in the report.
• Build and evaluate at least one suitable predictive or explanatory model (for example logistic regression, decision tree, or another well-reasoned choice). Compare alternatives where it strengthens the write-up.
• Provide readable notebooks / scripts with in-line commentary, clear visualisations, and repeatable runs so the marker can follow the workflow without tweaks.
• Write the accompanying report (approx. 2,000 – 2,500 words) that:
– explains and justifies each preprocessing decision
– describes the analytical methods and model selection
– interprets the statistical output and visual evidence in plain language
– cites recent, relevant academic literature in Harvard style to support the choices and discussion.

Deliverables
1. Cleaned dataset and any supplementary derived files
2. Colab notebook (plus optional WEKA files) with all code executed and cells neatly annotated
3. Final report in Word or PDF, ready to submit
4. Brief handover note so I can present or answer follow-up questions confidently.

Please keep variable names, column headers, and any code comments in English. I’m happy to answer clarifying questions quickly and can share the rubric and my current notebook as soon as we agree to proceed.