Solution to create a Classification Graph using Cloud for Internet of Health Applications from Evilasio Junior Research

Download this full repository

To add new Datasets:

• Open Server directory
• Open Datasets directory
• Open Source directory
• Open Datasets directory
• Add the new Dataset in the Datasets directory
• Back two directoris for Datasets directory
• Open DatasetManipulation directory
• Open Format directory
• Create code for preprocessing following the pattern of Dataset*.py files and Dataset*.py files
• Optional: Case need add new sensors or sesnsor combinations
• • Back for DatasetManipulation directory
  • Open Preprocessing directory
  • Create new preprocessing code file for new sensor compbination with same out signature of both other preprocessing code files in this directory
• Finaly Modify the MainFunctions.py to join the informations for new Dataset to the building graph instrunctions in this code file

• Python 3
• MySQL Server
• Flask
• TensorFLow 2.0
• Scikit Learn
• Numpy 1.* - Because Tensorflow

Run the following command to run the flask Web API (app.py) on server port 3000:

• flask run --host=0.0.0.0 --port=3000 &

Modify the ServerProcessAPI.py file to read the correct endpoint of your web API (e.g., api_url_base = '<API address>:3000')

Run "python MainServer.py" to run the server constant process or "python ServerProcessAPI.py" to test the web API

Informs the API that the graph must be updated:

• <API address>:3000/UpdateGraphRequest/2

Download the last updated complete graph:

• <API address>:3000/GraphRequest

Download the last generated optimized graph:

• <API address>:3000/OptimizeGraphRequest/download.xml

Request for optimization and download of the graph optimized by the application:

• <API address>:3000/OptimizeGraphRequest/<All types of sensors used by the application separated by an underscore> or
• <API address>:3000/OptimizeGraphRequest/<All types of sensors used by the application separated by an underscore>/<Threshold probability>

Request to download the trained models:

• <API address>:3000/OptimizeGraphRequest/saved_model/<Model Name>

Note: We suggest downloading the complete graph first to identify which models are trained and the types of sensors it uses. The complete Graph can also guide decision-making in the requirements elicitation and application design phases.