While descriptive analytics are sufficient for some insights , the predictions obtained from the historical data in future use cases is what's most useful for prescriptive analytics for the business. If there's an IOT sensor developer that's already worked to obtain an initial insight into their data, the next step usually involves a need for forecasting that can include some kind of supervised machine learning model( regression or classification) analysis. Machine learning as a service platforms (e.g. Azure) tend to be the preferred solution when your data and features are easy to model using a standard algorithm ( decision trees, regression trees) that's available on the cloud platform. But when you want a more custom solution we can come in and more exhaustively optimize the model's parameters to get you a better model for prediction.. We understand the problem with the limited availability of some feature sets that any group can incorporate into the model whether it's trading surveillance or predictive maintenance in operational scoring for continuous analysis.