/AAL-BayesianNetwork

Ambient Assisted Living - A Bayesian Network classification model for activity recognition in HAR research area

Primary LanguagePython

A Bayesian Network classification model for activity recognition in HAR research area

Aim of the project

Development of bayesian network that predict the action accomplished by a subject using the coordinates point of four accelerometers.

Dataset

  • Data acquired upon eight hour of activity
  • Four subject of different age and sex.
  • 165.633 samples

Networks created

picture picture

Obtained results

picture

EXECUTE CODE - AAL PROJECT

  1. Install conda environments (you can found it in folder Environments Conda; both Windows and MacOS version exists)

    • Launch python preprocessing_ugolino.py --> Preprocessing of data According to [W.Ugolino et al.] Or
    • Launch python preprocessing_our.py --> Preprocessing according to our version.

  2. Execute main.py

    • A bayesan network will be created using PGMPY library; evaluation of the accuracy of the model created, confusion matrix and metrics

  3. Execute model_pomegranate.py

    • A bayesan network will be created using Pomegranatelibrary; evaluation of the accuracy of the model created, confusion matrix and metrics

For every informations about preprocessing or function used to generate the networks look at Relazione.pdf or contact us.

Contact