STAT4011 Project Workspace

Two main parts:

• part1: Account classification experiments (ensemble strategies, feature extraction, model tuning). See subfolders (Qi Zihan, v1–v7) for detailed code and CSV outputs.
• part2: Crime analysis and time-series modeling (EDA, cleaning, visualization, Transformer/LSTM training). See part2/README.md for usage details.

Data policy

Large data directories are excluded from version control:

• part2/orginaldata/ (raw crime CSVs) — obtain from LAPD official releases or course Blackboard.
• part2/cleaned_data/ (processed outputs) — copy available at: https://drive.google.com/drive/folders/1I7GNt0kcznnNRe6aYDqoR17qqlPBmKNV?usp=sharing
• Virtual environment artifacts (**/bin/, **/lib/, **/include/, **/share/, pyvenv.cfg, .venv, venv/).

All processing and modeling code is included so datasets and outputs can be regenerated locally when raw data is available.