This project aims to make experiment for SDP much more easier. All the components are combined and examined the performance of every combinations.

• Python 3.5
• pip3

1. (Optional) virtualenv env -p python3 & source env/bin/activate
2. pip3 install -r requirements.txt
   • May have some errors while installing graphic related package on different system
3. cd src
4. python3 run.py

• data
  • Empty directory, synchronized data will be saved at here
• origin_data
  • Original source of the dataset
• report
  • Empty directory, generated report will be here
• src
  • Source code

• setting.py
  • Contains all the framework variables ( e.g. dataset path, selected dataset, selected methods, selected feature selection methods )
• run.py
  • Entry module that trigger everyother modules to start the process
• dataset.py
  • Handling dataset preprocessing
• feature.py
  • Contain functions that take feature and label as input and return the data with selected metrics depending on different implementations
• models.py
  • Core module of the framework that contain unsupervised methods for SDP. They are expected to take data as input, and output an one­dimensional array which classify every entities. 1 as defective and 0 as non­defective

• AEEEM
• NASA
• PROMISE