Unconstrained Logistic Regression

Model overview

Unconstrained logistic regression classifier using L1 (Lasso) regularization to predict taxonomy IDs from product tags. Unlike parameter-constrained models, this classifier uses ALL available tags and relies on L1 regularization to achieve sparsity by driving unimportant coefficients to zero.

5 CV folds

53.89% Mean accuracy

0.5694 Mean F1

0.25104309 Mean p-adic loss

4,258 Avg non-zero params

Total train samples: 29,328
Total test samples: 7,332
Number of tags (input features): 5,929
Avg non-zero parameters: 4,258 / 3,128,668 (99.9% sparse)
L1 regularization (C): 1.0000

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Non-zero params	Details
0	53.85%	0.5711	0.25790289	4,123	View
1	54.42%	0.5731	0.25738300	4,263	View
2	54.31%	0.5763	0.25506167	4,256	View
3	52.89%	0.5555	0.25225293	4,201	View
4	53.97%	0.5710	0.23261495	4,447	View