Does that mean there are no anomaly score values coming out of HTM.Java? Or that they are all 0.0? Or are you referring to the likelihood calculation you’re doing outside of HTM.Java?
The anomaly column is uniformly ‘none’. Im missing a param in the config somewhere.
I think we’re in #htm-java territory again, so I moved this topic.
I don’t run HTM.Java so I’m not sure what the problem is. I think you should be getting anomaly scores in your output if everything is set up correctly, so there is certainly a problem. Weren’t you getting these scores out previously?
Actually this is using the python samples. I was hoping to get better params by swarming over a sample of my data (~15K rows).
Sorry I’m totally confused because you’ve been running different platforms. Would it be possible for you to share your code and sample data set with me so I can try running it locally?
No worries - this has been a confusing journey so far.
Can you PM me an address? I can’t attach it here as it doesn’t accept non-image files.
@phil_d_cat, you could try posting a Gist?