README.md

---
annotations_creators:
  - expert-generated
language_creators:
  - found
language:
  - en
license:
  - cc0-1.0
multilinguality:
  - monolingual
pretty_name: Iris
size_categories:
  - n<1K
source_datasets:
  - original
task_categories:
  - tabular-classification
task_ids:
  - multi-class-classification
---

# Dataset Card for "iris"

## Dataset Description

- **Dataset Name**: Iris  
- **Dataset Type**: Tabular  
- **Source**: UCI Machine Learning Repository  
- **Languages**: English (feature labels)  
- **License**: Public Domain (freely usable)

### Dataset Summary

The Iris dataset is one of the most classic datasets in machine learning, often used for classification and clustering tasks. It contains 150 samples of iris flowers, each described by four features: sepal length, sepal width, petal length, and petal width. The task is to classify the samples into one of three species: *Iris setosa*, *Iris versicolor*, or *Iris virginica*.

This dataset is especially useful for:
- Supervised learning (classification)
- Unsupervised learning (clustering)
- Model explainability techniques
- Feature selection and dimensionality reduction

### Supported Tasks and Leaderboards

- **Classification**: Predict the species of iris based on the four numerical features.
- **Clustering**: Unsupervised grouping of samples into natural clusters.

### Languages

- The feature and label names are in English.

## Dataset Structure

### Data Fields

| Feature         | Type    | Description                             |
|----------------|---------|-----------------------------------------|
| sepal_length   | float32 | Sepal length in centimeters             |
| sepal_width    | float32 | Sepal width in centimeters              |
| petal_length   | float32 | Petal length in centimeters             |
| petal_width    | float32 | Petal width in centimeters              |
| label          | class label (str) | Species of the flower (`setosa`, `versicolor`, `virginica`) |

### Data Splits

There are no predefined splits, but you can randomly split the dataset for training and evaluation (e.g., 80/20 or 70/30).

### Example Row

```json
{
  "sepal_length": 5.1,
  "sepal_width": 3.5,
  "petal_length": 1.4,
  "petal_width": 0.2,
  "label": "setosa"
}