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

58.15% Mean accuracy

0.6206 Mean F1

0.154471 Mean p-adic loss

11,750 Avg effective params

Total train samples: 13,460
Total test samples: 3,365
Number of tags (input features): 1,627
Number of classes: 213
Avg tree depth: 346.6
Avg number of nodes: 1,520
Avg number of leaves: 761
Effective params formula: nodes × log₂(classes) = 1,520 × 7.73 ≈ 11,750

Cross-validation results

Fold	Accuracy	F1	P-adic loss (mean)	Tree depth	Nodes	Effective params
0	57.51%	0.6205	0.154315	358	1,535	11,873
1	57.04%	0.6046	0.136180	343	1,485	11,466
2	60.32%	0.6423	0.158625	345	1,483	11,461
3	55.86%	0.5958	0.176588	337	1,525	11,785
4	60.00%	0.6400	0.146645	350	1,573	12,167

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	ALLBIRDS::SILHOUETTE => DASHER	0.006498	2/5
2	SHOP: PARROT FOOD	0.006469	1/5
3	ALUMINIUM-FREE	0.006469	1/5
4	HEADWRAP	0.006469	4/5
5	SHOP: PELLETS	0.006469	3/5
6	TYPE_EXTENSION	0.006454	4/5
7	ORNAMENT	0.006454	5/5
8	HEALTH AND BEAUTY	0.006454	5/5
9	SIRUP	0.006454	5/5
10	WASHERS	0.006454	5/5
11	BUTTON	0.006454	5/5
12	PLATES	0.006454	5/5
13	POWER	0.006454	5/5
14	ADAPTER	0.006454	5/5
15	KEYS	0.006454	5/5
16	CERAMIC MAGNETS	0.006454	5/5
17	WIRE MANAGEMENT	0.006454	5/5
18	BOXERS	0.006454	5/5
19	LUXURY SOY CANDLE	0.006454	1/5
20	EXTENSION LADDERS	0.006454	1/5
21	YEAST	0.006436	4/5
22	YCRF_WOMENS-PERFORM-SHOES	0.006425	3/5
23	BRIEFS	0.006396	1/5
24	SHOP: ZUPREEM	0.006396	1/5
25	HEADSCARF	0.006396	1/5
26	POUCH	0.006286	4/5
27	WHISKEY/WHISKY	0.006172	5/5
28	JACKET	0.006150	5/5
29	EMOTO	0.006117	1/5
30	MOTORCYCLE	0.006086	4/5
31	CLOTHING	0.006082	5/5
32	DEODORANT	0.006025	4/5
33	LAUNDRY	0.006007	2/5
34	NON SLIP HEADBANDS	0.005910	4/5
35	TRUNKS	0.005901	4/5
36	CARPET	0.005855	5/5
37	GLOVES	0.005745	5/5
38	HEADBANDS	0.005707	1/5
39	COE_104	0.005702	4/5
40	VESTE	0.005681	1/5
41	CHRISTMAS	0.005607	5/5
42	ECOMM_PAGETYPE	0.005598	5/5
43	WALLET	0.005577	5/5
44	SS JACKET GILET	0.005566	5/5
45	CABINET BEDS	0.005475	5/5
46	PANTIES	0.005469	2/5
47	CABINET FACE PLATES	0.005412	5/5
48	RITUAL	0.005411	5/5
49	SANDALS	0.005389	4/5
50	FACE MISTS & SPRAYS	0.005374	5/5