th-ml-project

Tsinghua University - Machine learning final project

Architecture

Notebook for the project content only visualisation, results or run external script. Project is run by a python script.

• ml_project: Notebook for project
• dataset: Directory where dataset is store (test and train)
• download: Script for download data (no call in notebook)
• load: Script for load data (call in notebook)
• pickles: Pickles functions (cache function for optimize compute)
• preprocess: PreProcess functions
• script: Scripts operations directory (run script of, process script and debug script)
• utils: Functions utilities (like ThreadPool, ...)
• visualization: Function for visualization on notebook

---

Dependencies

• Python 3.5 >=
• Pip (for python 3.5 >=)
• Python packages describes on requirements.txt file

Usage

Install require package

pip install -r requirements.txt

Run

python scripts/compute_on_vm.py

Download dataset

Train

python3 download/train.py

Test

❗ Dateset is very big (> 6GB)

python3 download/test.py

Run Jupyter notebook (examples)

jupyter notebook ml_project.ipynb