How do I add predictive capabilities powered by custom machine learning models?
Custom machine learning automates decisions in businesses process by providing predictive capabilities or improving an application’s logic on data that already exists in your system. Unlike the cognitive services that are more oriented towards pre-built models for predefined use cases, custom machine learning models must be be trained and tested with actual specific business data. So, it is necessary to label and prepare your data, choose an algorithm, train the algorithm, tune it, and optimize it for deployment. Currently, there are solutions for creating machine learning models that require a minimum data science skills with a visual development language. One of these solutions is Azure ML. You can use OutSystems to integrate and leverage the application with Azure ML technology. Or, you can use OutSystems Machine Learning Builder.
With Azure ML, you can create an interactive, visual workspace that for easily and quickly building, testing, and deploying models using pre-built machine learning algorithms. No programming is required; the machine learning model is constructed by connecting datasets and analysis modules on an interactive canvas, and then deploying it.
Azure Machine Learning Studio publishes models as web services that can easily be consumed by OutSystems applications.
Components that connect to Azure ML are being created, and they will hide much of the complexity of creating, deploying and integrating these kinds of capabilities for main use cases.