Fold 3 - Understanding why coefficient 7159 was selected
Tag: AWARDS
Selected Coefficient: 7159
Taxonomy Path: 21.3.1
Taxonomy Name:
Products with this tag: 4
Unique residual values tried: 3
The algorithm tries each unique residual value as a candidate coefficient and selects the one with the lowest total p-adic loss. The selected row is highlighted.
| Candidate Value | Total P-adic Loss | Taxonomy Path | P-adic Expansion | Taxonomy Name | Selected? |
|---|---|---|---|---|---|
| 7159 | 0.012193 | 21.3.1 | 1.3.21 (6889 + 3*83 + 21) | Yes | |
| 270 | 0.012339 | 21.3 | 3.21 (3*83 + 21) | No | |
| 7491 | 0.036145 | 21.7.1 | 1.7.21 (6889 + 7*83 + 21) | No |
These are the training products that have the "AWARDS" tag (processing sequence 4722). The residual is the encoded taxonomy value minus all coefficients from previously processed tags.
| Product ID | Title | Residual Before | Taxonomy Before | Tags Already Processed | Residual After | Taxonomy After | Tags Not Yet Processed |
|---|---|---|---|---|---|---|---|
| 21816 | Best Firms To Work For Award | 7491 | none | 332 | none | ||
| 21834 | 2023 Marketing Excellence Award | 7159 | none | 0 | none | ||
| 21844 | 2023 Trifecta Award Bundle | 7159 | none | 0 | none | ||
| 21841 | Top New Venture Award | 270 | none | -6889 | none |