- Our ultimate goal is to build a non-exclusive end-to-end IoT device predictive maintenance and management solution, from on-field devices all the way to a web console. The solution should be easy-to-use, flexible for all devices, and scalable to accommodate future growth.
- http://www.devxia.com/crouching_tigers
- Please use http instead of https. AWS EC2 does not provide SSL certificates.
report.pdf
code/iot_project.ipynb
: the main code for our project, including modules for:- Data ETL/EDA
- Model Training
- Model Testing
- Anomaly Detection Fitting/Testing
- Plot
code/project_iot.py
: old models using pyspark machine learning pipeline and DecisionTree classification algorithm for the project. Discarded in the latest modulescode/LinearRegression.py/LogisticRegression.py/options_regression.py
: similar to project_iot.py, but using different regression algorithmscode/AnomalyDetection.ipynb
: Test code for anomaly detection using different datasetcode/FeatureSelection.ipynb
: Test code for feature selection using different datasetmodel
: saved model exported from Tensorflow, deployed on Google Cloud Machine Learning Engine.dataset/demo_data.zip
: The raw dataset we actually used for final result.web/node_server
: Proof of concept Node.js server that simulate on-field device uptime and heartbeat- See more instruction at
web/node_server/README.md
- Deployed on AWS
- Example REST API endpoint:
GET
http://ec2-18-236-179-13.us-west-2.compute.amazonaws.com:5000/machines
- See more instruction at
web/frontend
: Proof of concept React frontend dashboard- See more instruction at
web/frontend/README.md
- Deployed on http://www.devxia.com/crouching_tigers
- Template starter by Creative Tim (https://www.creative-tim.com)
- Commnunicate with model deployed on Google Cloud Machine Learning Engine and Node.js server on AWS.
- See more instruction at