Describe the Wine Groups

A cluster number means nothing until someone can say what the group is. A wine merchant wants each group described in real units.

Task: write describe_groups(k, columns).

• Load the wine data with load_wine(return_X_y=True, as_frame=True).
• Scale all 13 columns with StandardScaler().fit_transform(X).
• Fit KMeans(n_clusters=k, random_state=0, n_init=10) on the scaled data.
• For each cluster, find the mean of each name in columns, using the original, unscaled values.
• Return a list with one dict per cluster. Each dict has "size" (the number of wines in it, a plain int) and one key per column name, holding that mean as a plain float rounded to 2 decimal places.
• Order the dicts from the largest cluster to the smallest. No two clusters have the same size.