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

50.90% Mean accuracy

0.5431 Mean F1

0.217154 Mean p-adic loss

64,105 Avg effective params

Total train samples: 54,236
Total test samples: 13,559
Number of tags (input features): 4,981
Number of classes: 669
Avg tree depth: 1154.2
Avg number of nodes: 6,829
Avg number of leaves: 3,415
Effective params formula: nodes × log₂(classes) = 6,829 × 9.39 ≈ 64,105

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Tree depth	Nodes	Effective params
0	50.16%	0.5332	0.221047	1149	6,851	64,303
1	51.31%	0.5478	0.215898	1146	6,885	64,592
2	51.67%	0.5476	0.228083	1152	6,833	64,089
3	50.86%	0.5470	0.227150	1159	6,781	63,675
4	50.50%	0.5402	0.193594	1165	6,797	63,869

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	ACCESSORIES	0.004623	5/5
2	NEW	0.003993	5/5
3	WOMENS	0.003007	5/5
4	BIRTHDAY	0.002948	5/5
5	WOMEN	0.002793	5/5
6	KIDS	0.002629	5/5
7	RETAIL	0.002604	5/5
8	HAT	0.002601	5/5
9	BLACK	0.002502	5/5
10	MAKEUP	0.002452	5/5
11	BRUSH	0.002447	5/5
12	CLOTHING	0.002283	5/5
13	CAR COASTERS	0.002126	2/5
14	SHOP: PARROT FOOD	0.002124	1/5
15	COOKING OIL	0.002122	1/5
16	SHOP: ZUPREEM	0.002122	1/5
17	FOOD COLOR	0.002119	4/5
18	DRESS	0.002117	5/5
19	PRINTED SILK	0.002117	3/5
20	ALL30MLPRODUCTS	0.002115	4/5
21	_RELATED_HOT-CEREAL-LANDING-PAGE	0.002115	4/5
22	BICYCLES AND MOTORCYCLES	0.002114	2/5
23	SIRUP	0.002112	5/5
24	TOWEL-RAILS	0.002112	5/5
25	THERMOSTATS	0.002112	5/5
26	ELECTRIC-SKATEBOARDS	0.002112	5/5
27	INKYBAY_CUSTOM	0.002112	5/5
28	HEALTH AND BEAUTY	0.002112	5/5
29	IMT TOOLS	0.002112	5/5
30	RAFCAMERA	0.002112	5/5
31	CHEESE TOOL	0.002112	5/5
32	PLANT STANDS AND BENCHES	0.002112	5/5
33	KEYS	0.002112	5/5
34	WARRANTY-SEATS_2	0.002112	5/5
35	MOTORCYCLE RAMP	0.002110	3/5
36	CAR ACCESSORIES	0.002104	2/5
37	PRINTED POCKET SQ	0.002104	2/5
38	SHOP: PELLETS	0.002104	3/5
39	PALEO HOT CEREAL	0.002097	1/5
40	NICOTINE	0.002097	1/5
41	CATEGORY_BATHMATS	0.002085	5/5
42	_RELATED_GRANOLA-LANDING-PAGE	0.002051	5/5
43	KNITTED BEAN BAG	0.002046	1/5
44	BEANBAG	0.002025	1/5
45	WIRE MANAGEMENT	0.002017	5/5
46	2024	0.002011	3/5
47	13 MONTHS TO 24 MONTHS	0.002003	2/5
48	INTERCHANGEABLE WATCH STRAP	0.001995	2/5
49	SUPER CHUNKY BEAN BAG	0.001995	1/5
50	HALL50OFF	0.001991	2/5