HAR-bayesian-network

The project consists in creating a bayesian network model in order to predict the action of a person, basing on the given training set which consists in 3D acceleration vectors for each sensor analyzed in the the project Wearable computing: Accelerometers' data classification of body postures and movements

Dependencies

You can install the dependencies using pip:

pip install -r dependencies.txt

Inference

Make a sequence of inferences writing them in the input path as array of strings. Results will be printed in the output path (configurable in configs/Config.py). Following default paths:
input: inference/in.txt
output: inference/out.txt

python inference.py

Make a single custom inference from the given model and return the class-action, for example

python inference.py "x1=-1;y1=100;z1=-97;x2=4;y2=85;z2=-123;x3=24;y3=98;z3=-94;x4=-210;y4=-87;z4=-162"

Generate the model

Generate dynamically the model of the Bayesian Network (skeleton and CPDs)

python skeleton.py

Get accuracy

Stimating the accuracy of the given model (based on 10% of the dataset)

python test.py

Authors

• Federico Bottoni - FedericoBottoni

• Mattia Artifoni - m-artifoni

• Luca Brena - lbrena2

License

HAR-bayesian-network source code is licensed under the MIT License.