make_animals_check_data: Gnostic Animals Dataset

The make_animals_check_data function retrieves the 'Animals' dataset (Rousseeuw & Leroy, 1987), a classic small and challenging dataset for robust regression. It contains body weight and brain weight for 28 animals.

---

Overview

This dataset is famous in robust statistics because it contains severe outliers that can skew traditional regression models. The goal is typically to model the relationship between body weight and brain weight.

• Features (\(X\)): Body weight in kilograms (kg).
• Target (\(y\)): Brain weight in grams (g).
• Challenge: The dataset spans several orders of magnitude (from a Mouse to a Brachiosaurus). It includes 3 dinosaurs (Diplodocus, Brachiosaurus, Triceratops) which act as "bad leverage" points—they have massive body weights but relatively small brains compared to the mammalian trend.
• Suggested Usage: Regression analysis, often with log-transformation: \(\log(\text{Brain}) \sim \log(\text{Body})\).

---

Returns

Return	Type	Description
X	numpy.ndarray	Body weight in kg. Shape (28, 1).
y	numpy.ndarray	Brain weight in g. Shape (28,).
names	list of str	The common name of the animal for each data point (e.g., "Mouse").

---

Example Usage

from machinegnostics.datasets import make_animals_check_data
import numpy as np

# Load the dataset
X, y, names = make_animals_check_data()

# inspect heaviest animal
heaviest_idx = np.argmax(X)
print(f"Heaviest: {names[heaviest_idx]} ({np.max(X)} kg)")
# Output: Heaviest: Brachiosaurus (87000.0 kg)

# Example: Inspecting first 5 entries
for i in range(5):
    print(f"{names[i]}: Body={X[i,0]}kg, Brain={y[i]}g")

---

Dataset Content

The dataset includes the following animals: Mountain beaver, Cow, Grey wolf, Goat, Guinea pig, Diplodocus, Asian elephant, Rhesus monkey, Kangaroo, Golden hamster, Mouse, Rabbit, Sheep, Jaguar, Chimpanzee, Rat, Brachiosaurus, Mole, Pig, Echidna, Triceratops, Pigmy marmoset, African elephant, Human, Potar monkey, Cat, Giraffe, Gorilla.

---

Source: Rousseeuw, P. J., & Leroy, A. M. (1987). Robust Regression and Outlier Detection. John Wiley & Sons.

Author: Nirmal Parmar