Decision Tree Classifier

Model overview

Decision tree classifier for predicting taxonomy IDs from product tags. Unlike parameter-constrained models, this classifier uses ALL available tags. The tree structure provides natural interpretability - each decision node tests whether a specific tag is present, leading to a taxonomy prediction at the leaves.

5 CV folds

55.94% Mean accuracy

0.5940 Mean F1

0.187412 Mean p-adic loss

26,232 Avg effective params

Total train samples: 24,724
Total test samples: 6,181
Number of tags (input features): 2,797
Number of classes: 368
Avg tree depth: 606.0
Avg number of nodes: 3,080
Avg number of leaves: 1,540
Effective params formula: nodes × log₂(classes) = 3,080 × 8.52 ≈ 26,232

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Tree depth	Nodes	Effective params
0	55.82%	0.5976	0.204143	593	3,047	25,971
1	54.66%	0.5860	0.194684	592	3,069	26,123
2	58.44%	0.6169	0.141027	606	3,171	27,016
3	55.59%	0.5858	0.195708	628	3,095	26,356
4	55.20%	0.5838	0.201496	611	3,017	25,692

Feature importances (Top 50)

Feature importances are averaged across all 5 folds. The "Folds" column shows how many folds used this feature.

Rank	Tag	Avg Importance	Folds
1	WOMEN	0.003825	5/5
2	ACCESSORIES	0.003810	5/5
3	SHOP: ZUPREEM	0.003780	1/5
4	EMOTO	0.003780	1/5
5	BEDROOM FURNITURE	0.003751	2/5
6	HEADSCARF	0.003751	2/5
7	NICOTINE	0.003742	2/5
8	HANDSCULPTED	0.003740	1/5
9	MELISSA-DOUG	0.003739	2/5
10	YCRF_WOMENS-PERFORM-SHOES	0.003739	2/5
11	AFRICAN BASKET	0.003737	5/5
12	BUTTON	0.003737	5/5
13	PLATES	0.003737	5/5
14	THERMOSTATS	0.003737	5/5
15	WARRANTY-SEATS_2	0.003737	5/5
16	KEYS	0.003737	5/5
17	GARMIN BIKE MOUNTS	0.003737	5/5
18	HEALTH AND BEAUTY	0.003737	5/5
19	POWER	0.003737	5/5
20	SIRUP	0.003737	5/5
21	SMALL GIFTS	0.003737	5/5
22	YEAST	0.003737	5/5
23	PLANT STANDS AND BENCHES	0.003737	5/5
24	CERAMIC MAGNETS	0.003736	4/5
25	OVER 25	0.003735	3/5
26	ALLBIRDS::SILHOUETTE => DASHER	0.003735	3/5
27	ALL30MLPRODUCTS	0.003733	3/5
28	BED2.0	0.003727	3/5
29	HEADWRAP	0.003727	3/5
30	DURSTON FLAT NOSE	0.003722	1/5
31	SHOP: PELLETS	0.003720	3/5
32	ALUMINIUM-FREE	0.003697	1/5
33	SLIPPERS	0.003675	5/5
34	PLATE	0.003648	4/5
35	OUTDOORS	0.003645	4/5
36	POUCH	0.003639	4/5
37	CANVAS	0.003637	4/5
38	PICTURE	0.003573	4/5
39	WHISKEY/WHISKY	0.003563	5/5
40	25 MONTHS TO 36 MONTHS	0.003556	3/5
41	OVER-30	0.003556	5/5
42	WIRE MANAGEMENT	0.003542	5/5
43	DEODORANT	0.003541	4/5
44	SHOP	0.003523	5/5
45	RESPIRATORY	0.003509	2/5
46	TYPE_EXTENSION	0.003506	2/5
47	MOTORCYCLES	0.003491	1/5
48	MOTORCYCLE	0.003476	1/5
49	GLITTER	0.003468	5/5
50	ORNAMENT	0.003465	5/5