Firstly, we trained a machine learning model on Amazon Machine Learning service with the criterions described above. Secondly, we use Amazon Web Service for data storage and transferring. Data flow is shown as the figure above. Intel Edison works as a gateway to the AWS: after gathering data, it sends data to AWS Kinesis. Then, AWS Lambda Function fetches data from Kinesis, and fit it to the machine learning model given by Amazon Machine Learning Service. Finally, SNS will obtain the prediction result and send it to a Flask server, which is implemented on EC2.