Skip to content

Attributes

Piotr Tylczyński edited this page Apr 19, 2023 · 3 revisions

Attributes

During each phase of benchmarking there is different subset of available Attributes to use. User is free to change them as benchmark goes.

Attributes per phase

Phase Attributes available
Preprocessing RUNS, X_RAW, Y_RAW, X, Y, STORE
Init Model RUNS, X_RAW, Y_RAW, X, Y, MODEL, STORE
Train X (selected), Y (selected), MODEL, STORE
Test X (selected), Y (selected), MODEL, STORE
Evaluate X (selected), Y (selected), Y_TRUE, MODEL, STORE
Summarize X (all), Y (all), MODEL, STORE

Attributes usage

Attributes are implemented via simple Python map. This mean user can change them during runtime, however doing this reckless can cause falsified results.

To access attributes in any phase just use:

attributes[Attributes.X] = ....

Attributes meaning

RUNS - number of folds that will be provided for train and test

X_RAW - raw X data read from .caddo file

Y_RAW - raw Y data read from .caddo file

X - processed X data

Y - processed Y data

MODEL - model to train and test

Y_TRUE - this is subset of Y provided exclusively to test function to compare it with Y predicted

STORE - this is python map() object, you can put there whatever you want and it will be passed through each step intact

Clone this wiki locally