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

50.45% Mean accuracy

0.5481 Mean F1

0.24156681 Mean p-adic loss

4,552 Avg non-zero params

Total train samples: 36,440
Total test samples: 9,110
Number of tags (input features): 3,664
Avg non-zero parameters: 4,552 / 1,793,651 (99.7% sparse)
L1 regularization (C): 1.0000

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Non-zero params	Details
0	51.64%	0.5557	0.24035637	4,541	View
1	48.40%	0.5305	0.25629307	4,666	View
2	51.72%	0.5630	0.22957934	4,513	View
3	49.76%	0.5374	0.24944502	4,430	View
4	50.73%	0.5538	0.23216024	4,610	View