The Cloud Academy team shows how to build a sentiment analysis machine learning model by using a pubic dataset and how to deploy it to production with AWS Lambda and API Gateway.

OS libraries can be installed as follows (Linux):

$ sudo yum install -y atlas-devel atlas-sse3-devel blas-devel gcc gcc-c++ lapack-devel python27-devel

Python requirements can be installed with pip.

$ pip install -r requirements.txt

You can train and persist the model into file by executing:

$ python main.py

You can run the model via CLI as follows:

$ python predict.py "This function is awesome"
> positive

Simply execute the following deployment script:

$ ./deploy.sh

It will automatically create a new deployment package containing every Python dependency and the OS libraries required by scikit-learn and numpy, together with the "lambda" folder.

Here you can find the whole stack ready to be uploaded (reference).

Once the deployment package is ready, you can create a new AWS Lambda function and bind it to a new Amazon API Gateway endpoint.

• Cloud Academy & AWS: how we use Amazon Web Services for machine learning and data collection (webinar)

• Using Scikit-Learn In AWS Lambda (article by Ryan Brown)

• Sentiment Analysis dataset (by cs.stanford.edu)

• Amazon Serverless Architecture (Cloud Academy blog)

• Google Cloud Functions VS AWS Lambda (Cloud Academy blog)