Remove loom from the build.#219
Open
LeifAndersen wants to merge 41 commits into
Open
Conversation
added 30 commits
August 5, 2025 13:02
1. Does NOT use loom. Loom seems to be broken at the moment. 2. Requires the model name to be '0.edn`, not `sample.0.edn`. 3. Is only tested in one container.
Yes its non-standard, but our pipeline produces it.
They aren't valid json.
It requires nixos to run, which is an unrealistic requirement.
In that case we can't learn anything, but that shouldn't bring the pipeline to a halt.
(Next commit will test this out.)
Still working on proper tests.
I logged it, and switched the dependency to pivoted.csv
The pivot script handles schemas for unpivoted data. Putting it there throws off the `AST` step.
A println inserted in the script was the problem. Fixed now.
This premodel (or shadow model) is not as accurate as the real one, but the key is that its significantly faster to compute.
While the onehotencoder is better, it doesn't work with the older version of scikit learn used in the image. So for this simple purpose, we can get away with using get_dummies directly.
Before when we did dropna the index changes completely through off the result. Now we're dropping the NA rows entirely.
added 11 commits
September 12, 2025 13:09
We were filtering out the pre-pivoted schema, if we filter out the post-pivot one, the guessing process seems to be up to 50% faster.
The implementation for guess_schema is forthcoming.
On the upside, its a lot faster, but is not as robust as the clojure method.
Otherwise we can't call the munge function.
When requested, the python predictor can now shortcircuit the clojure schema predictor, taking times from minute (or hours) down to just a few seconds.
They were wrong. nominal -> numerical categorical -> nominal
Thee ntire data structure was being converted to a string before being emited, this caused an out of memory error on large (bigger than 1 GB) data structures. We can, however, simply write out the structure to the file, rather than first converting it to a string in memory.
While ensemble/ensemble is usually better, in this case we don't need the extra safety checks, additionally larger models will cause the checks to run out of memory.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Since it doesn't seem to work.