Amazon SageMaker is a serverless service offered by AWS that allows ML experts and data scientists to build ML models. Similar to how transcribe, polly, Rekognition, etc, all use pre-built ML models to analyze a specific type of data except that SageMaker allows you to make your own ML model. First, you accept a range of data. Then, you label this data and specify which values are to be used by your ML model and what these values correspond to. Next, you train and tune your ML model based on this data. Finally, when new data comes in, you can apply this ML model that has been tuned from the data provided to it in order to analyze and make predictions on the provided information.
Forecast is a very simple service, it does exactly what the name says. Forecast provides a “forecast” of collected data to determine with, high accuracy, where the forecast is going to lead. For example, you may have a forecast of sales. You would provide all the information relating to your sales, the more information the more accurate the forecast, upload it to S3, and then start the forecast service which will generate a forecast model and give you an accurate sales forecast. This can greatly decrease the time it takes to perform forecasting, reduces time for financial planning, and is far more accurate than manually reviewing data.
Kendra is a document search service that uses ML to analyze numerous sources of data to provide search functionality with natural language search capabilities, think similar to google. It can learn from user interactions and feedback to create more fine-tuned results, incremental learning. This service could be very useful in an employee intranet where everyone needs the ability to easily search for information across documentation without having to go through every data source manually.