Hi, my name is Kenneth, I am working with Ignition Machine Learning Manager and I have historical tag data (temperature and humidity) collected over several months.
My goal is to build a model that can predict future values (for example, predicting temperature 10 minutes ahead) based on current and past data.
Currently, I am able to:
However, I am unsure about the correct way to structure the dataset for time-based predictions.
So I am working and creating models with this Ignition Machine Learning Manager What type of algorithm can I use to train a model entirely within Ignition, without depending on external systems?
What I need is to create a trained model that can be read directly by Ignition or through a tag, and use it to visualize or predict values some time into the future—for example, 10 minutes ahead of real-time data.
I believe this could be achieved using the Machine Learning Manager by creating and deploying a model within Ignition itself. Could you please guide me on the best approach or recommended algorithms for this use case?
Any help would be greatly appreciated. Thank you in advance.
kennethandradeom, I echo your thoughts above. Has anyone out there had any success with the Machine Learning Manager? I have spent the last couple months attempting to get just a single model that works with no success and have spent hours and hours on the phone with technical support. Based on my experience, there are many issues that need resolved with this Perspective project. As much as I would like to use the Machine Learning Manager, I am feeling I may need to look elsewhere given the amount of time I have spent with this and still have no real results yet. The person in technical support is trying their hardest to help me out, but I feel they lack the training they need to troubleshoot this project. It is difficult for this person to troubleshoot the project, when I am telling him how I think the project should work. The developer of the Machine Learning Manager has been helpful, but it takes many days if not over week to get a response or project fix.
The documentation specific to the project is limited, but does give many links to generic resources. While this is somewhat helpful, it doesn't help out with specific issues related to the Machine Learning Manager.
Again, if anyone outside of Inductive Automation has had a positive experience with the Machine Learning Manager, I would like to hear more about your experience.
