Great question. In this “golden age of IOT” you have many choices. Here are a couple I have looked at.
I am a big fan of Ubidots but other platforms have similar features. You could send your data to Ubidots and then use the IBM Watson plug-in to do the predictive part. Then, you could build a dashboard that would show actual and predicted results over time. Finally, you can set alert events if your predictions (or actual) data crosses thresholds. These services are set up to reduce the amount of coding you need to do in order to get up and running.
Another popular choice is AWS Greengrass which is coupled with their IOT services. In my exerpience, the solutions are more flexible but may require more expertise and configuration on your end.
Well, two things. First, I don't think that it would be in Particle's wheelhouse to build a tutorial on predictive methods. Instead, they provide the platform to sense and respond to data. Other companies such as Google, Amazon, IBM and others will build tools that take the data collected by the Particle device and apply predictive models. So, I think you have to look at companies like this for a tutorial.
For example, IBM offers a machine leanring desktop eBook and demo tooling. Perhaps this would be a great place to start. There may be other, better, resources so perhaps someone else in the community can provide links as well:
I also found this post which, while not about prediction in particular, brings a lite version of Google's popular machine leanring platform, Tensor Flow, to a 3rd Generation device.