Results of this run
Growing Neural Gas on Wine (UCI wine recognition) grows to maxNodes 10. Gas silhouette is 0.118, gas Calinski-Harabasz is 29.9, gas Davies-Bouldin is 1.844. Held-out wine_class versus gas units: adjusted Rand 0.316. 0 noise rows out of 178 far from a unit. Gas unit 0 on Wine (UCI wine recognition) holds 21 rows (proline 1125.2 / magnesium 107.3). Gas unit 1 on Wine (UCI wine recognition) holds 23 rows (proline 657.0 / magnesium 96.57). Gas unit 2 on Wine (UCI wine recognition) holds 18 rows (proline 488.7 / magnesium 86.39). Gas unit 3 on Wine (UCI wine recognition) holds 11 rows (proline 916.6 / magnesium 110.2). Gas unit 4 on Wine (UCI wine recognition) holds 10 rows (proline 632.6 / magnesium 112.3). Gas unit 5 on Wine (UCI wine recognition) holds 18 rows (proline 469.7 / magnesium 89.44). Gas unit 6 on Wine (UCI wine recognition) holds 11 rows (proline 585.0 / magnesium 111.3). Gas unit 7 on Wine (UCI wine recognition) holds 28 rows (proline 1174.7 / magnesium 106.8). Gas unit 8 on Wine (UCI wine recognition) holds 21 rows (proline 610.8 / magnesium 95.86). Gas unit 9 on Wine (UCI wine recognition) holds 17 rows (proline 493.8 / magnesium 91.35). Node-budget path: max_nodes 6 grows 0.202 with 6 groups on Wine (UCI wine recognition); max_nodes 10 grows 0.118 with 10 groups on Wine (UCI wine recognition); max_nodes 16 grows 0.095 with 16 groups on Wine (UCI wine recognition). Gas scatter keeps 55.4%.

| max_nodes | Silhouette | Clusters |
|---|---|---|
| 6 | 0.202 | 6 |
| 10 | 0.118 | 10 |
| 16 | 0.095 | 16 |
# Product Growing Neural Gas: max_nodes=10, random_state=42.
# Features are StandardScaler-transformed before the fit.
Who, how, and why
Method: StandardScaler, drop constant columns. Code version 3e4afc4b4352 (scripts/seo/build_explore_artifacts.py). scikit-learn 1.4.2. Seed 42. Dataset source: UCI Machine Learning Repository, wine recognition. License: CC BY 4.0. The wine_class column was held out of the fit and used only for the adjusted Rand score.
Results computed on 2026-11-02.
The results paragraph is filled from these numbers by a fixed template in the site code, so the scores can be checked against the same dataset in the tool.
Maintainer: nikolaou1982.
Contact: nick1982100@gmail.com.
