---

license: cc0-1.0

Iris Species Dataset

The Iris dataset is a classic dataset in machine learning, originally published by Ronald Fisher. It contains 150 instances of iris flowers, each described by four features (sepal length, sepal width, petal length, and petal width), along with the corresponding species label (setosa, versicolor, or virginica).

It is commonly used as an introductory dataset for classification tasks and for demonstrating basic data exploration and model training workflows.

Supported Tasks and Leaderboards

• Multi-class Classification: Predict the species of the iris flower based on its four numeric features.

Dataset Structure

The features are:

• sepal_length (float): Length of the sepal in centimeters
• sepal_width (float): Width of the sepal in centimeters
• petal_length (float): Length of the petal in centimeters
• petal_width (float): Width of the petal in centimeters

The target variable is species (string). Labels ares:

• setosa
• versicolor
• virginica

This dataset traditionally comes as a single collection of 150 rows.

Background

The Iris dataset was first introduced by Ronald Fisher in 1936 and later became known through UCI Machine Learning Repository. This version was obtained from the UCI Machine Learning Repository's Kaggle

Source Data:

• Repository: UCI Machine Learning Repository
• Original Paper: R. A. Fisher (1936). “The Use of Multiple Measurements in Taxonomic Problems.” Annals of Eugenics 7 (2): 179–188.

Citation

Fisher’s original publication:

@article{fisher1936use,
  title={The use of multiple measurements in taxonomic problems},
  author={Fisher, Ronald Aylmer},
  journal={Annals of eugenics},
  volume={7},
  number={2},
  pages={179--188},
  year={1936},
  publisher={Wiley Online Library}
}

Usage Example

Here’s a minimal example in Python using the datasets library:

from datasets import load_dataset

# This dataset is hosted in Hugging Face Hub under "brjapon/iris-dataset"
dataset = load_dataset("brjapon/iris")

# The dataset might contain a default split or just one split
df = dataset["train"].to_pandas()

print(df.head())

Or using Scikit-Learn’s built-in functionality:

from sklearn.datasets import load_iris

iris = load_iris()
X = iris.data  # shape (150, 4)
y = iris.target  # shape (150,)

Limitations and Potential Bias

The dataset contains only three iris species from a limited geographic region (the Gaspe Peninsula in Canada). It is quite small (150 samples) and was originally collected for a statistical illustration rather than real-world machine learning applications. It serves primarily as a simple demonstration dataset and is not representative of broader botanical diversity.