**The most important metric is `petal_length`.** It best separates the three species, which is what this dataset is for. This is a judgement based on how far apart the species sit on each measurement, not a statistical test. **Why petal length** - The species averages are far apart: setosa 1.46, versicolor 4.26 and virginica 5.55. - Setosa's range (1.0–1.9) does not overlap the other two at all. Versicolor runs 3.0–5.1 and virginica 4.5–6.9, so only those two overlap, in a narrow 4.5–5.1 band. - Sepal width is a much weaker separator. The species ranges overlap heavily (setosa 2.3–4.4, versicolor 2.0–3.4, virginica 2.2–3.8), and the averages (3.43, 2.77 and 2.97) are close together. - Petal width behaves much like petal length, with averages of 0.25, 1.33 and 2.03. It is a close runner-up, and in practice you would use the two together. **Highest and lowest records for petal length** - **Highest:** a *virginica* specimen with petal_length 6.9 (petal_width 2.3, sepal_length 7.7, sepal_width 2.6). - **Lowest:** a *setosa* specimen with petal_length 1.0 (petal_width 0.2, sepal_length 4.6, sepal_width 3.6). These two records are also the species-level extremes: 6.9 is virginica's gold-table maximum and 1.0 is setosa's minimum. The shortest and longest petals therefore come from the two species that are furthest apart, and the metric spans about a sevenfold range. The per-species figures come from the `iris_by_species` gold table. The extreme records come from a query on `raw.iris` that returned one row each for the highest and lowest petal length.