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

58.39% Mean accuracy

0.6199 Mean F1

0.27466602 Mean p-adic loss

2,663 Avg non-zero params

Total train samples: 18,320
Total test samples: 4,580
Number of tags (input features): 6,936
Avg non-zero parameters: 2,663 / 2,421,013 (99.9% sparse)
L1 regularization (C): 1.0000

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Non-zero params	Details
0	58.37%	0.6140	0.27661140	2,654	View
1	55.79%	0.5932	0.30813295	2,651	View
2	59.91%	0.6457	0.27618476	2,608	View
3	58.71%	0.6225	0.25549024	2,696	View
4	59.16%	0.6240	0.25691074	2,706	View