One of the coolest things I have seen this last year has been the release of Azure Machine Learning from Microsoft. http://azure.microsoft.com/en-us/services/machine-learning/ Demos of the new service have been making the rounds in PASS and Dot Net users groups.
If you haven’t seen it, the best way I can describe it is predictive modeling using a GUI similar to SSIS or Windows Workflow. Using it follows a pretty standard cycle.
- Upload data to analyze
- Create an analytical model using pre-defined algorithms or provide your own custom models
- Exercise a percentage of your data to be executed through the models and train it to make predictions
- Exercise a percentage of your data through the machine determined path to determine accuracy of the learning
- Deploy your model into a production environment for future analysis and prediction
- Wash/rinse and repeat (in other words, you need to evaluate your model in an ongoing basis because the predictive accuracy may change over time)
You can do this with many other modeling tools. R is a popular programming language for those writing predictive programming. So, what makes Azure ML unique? It is a tool that may be used by non-programmers to great effect. It is a drag and drop workflow like painter. There is a learning curve; but it is nowhere near as big as learning to program in R.
Ok, so you already know how to program in R. You like Python to plug things together. These tools are also enabled in Azure Machine Learning, and can be incorporated into workflows. You workflow can use custom R routines in conjunction with routines already built into Azure ML.
This is a product worth your time to at least become acquainted. Take a look and share your thoughts in our comments.
Cheers,
Ben