MSBD-5002-project

Github: here

Prerequisite

• conda, either through miniconda or miniforge (osx-arm64)
• Git LFS
  MacOS: brew install git-lfs && git lfs install
  Others: See here

Getting Started

1. Note: the following files inside data-sets/KPI are too large to be committed, please place them in manually.

   • test-data.hdf
   • train-data.csv

2. cd into project directory, and create conda env

   conda env create -f environment.yml
   # as specified in the yml, new 
   # env will be named `5002-project`
   conda activate 5002-project

   Note: should you encounter any issues, you can also try installing dependencies with pip

   pip install -r requirements.txt

3. Confirm python version is correct

   python --version
   # Python 3.9.x

4. cd into src/ folder

5. Run the main application. A progress bar should show up, and interactive graphs (.html) should start appearing in the /{projectdir}/output/ folder.

   python main.py
   #  0%|         | 0/250 [00:00<?, ?it/s]

6. Upon completion, a submission.csv file shall appear in the /{projectdir}/output/ folder.

Managing Conda + Pip Environment

• Create conda env: conda env create -f environment.yml
• Activate: conda activate 5002-project
• Install pip packages: pip install -r requirements.txt
• Export conda package list: conda env export --no-builds --from-history > environment.yml
• Export pip package list: pip list --format=freeze > requirements.txt

Reference

• Matrix Profiles Time Series Mining