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

52.51% Mean accuracy

0.5566 Mean F1

0.25519512 Mean p-adic loss

4,473 Avg non-zero params

Total train samples: 30,760
Total test samples: 7,690
Number of tags (input features): 5,904
Avg non-zero parameters: 4,473 / 3,248,931 (99.9% sparse)
L1 regularization (C): 1.0000

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Non-zero params	Details
0	50.92%	0.5467	0.26165793	4,328	View
1	53.69%	0.5716	0.24380151	4,561	View
2	53.43%	0.5624	0.25937062	4,447	View
3	53.02%	0.5556	0.26457591	4,388	View
4	51.50%	0.5466	0.24656964	4,639	View