Visualise positive, negative, and uncertain examples

[MatthewJA]

Before you go down the route of finding features, visualise the IR and radio images of the positive examples that are classified negative by your predictor. 5-10 image patches from:

score approximately -10
score approximately 0
And for comparison, look at 5-10 patches where the score is >5.

At the same time, show the flux values (all other non-image features).

[MatthewJA]

Still working on this issue, but I'd like to mention that the most negative scored positive example is this:

I see why.

[MatthewJA]

Here's the five lowest-scored positive examples.

By the way, looked up WISE on the most negative one — looks like it has no corresponding IR, so the Norris-derived label is probably wrong.

[MatthewJA]

Top row are least positive, bottom row are the least negative. Not sure of a good way to get the infrared from WISE, since I'm having trouble figuring out the API.

[MatthewJA]

And finally, the bottom row is the 5 most certain (with scores between 30 and 50).