What is the Iris Dataset dataset?

Fisher's Iris dataset: 150 flower samples across 3 species with sepal/petal measurements. The classic benchmark for classification algorithms and machine learning education since 1936.

How can I download the Iris Dataset dataset?

You can download the Iris Dataset dataset in CSV or Parquet format directly from this page. Each table has its own download buttons.

What format is the Iris Dataset dataset available in?

This dataset is available in CSV and Parquet formats. CSV is great for spreadsheet applications, while Parquet is optimized for data analysis tools like Pandas and DuckDB.

What is the license for the Iris Dataset dataset?

This dataset is available under the CC BY 4.0 license. See the full license at https://creativecommons.org/licenses/by/4.0/

How many tables are in the Iris Dataset dataset?

The Iris Dataset dataset contains 1 table: iris.

Iris Dataset

Last updated 2 weeks ago•December 27, 2025

Source:UCI Machine Learning Repository

License:CC BY 4.0

Version:1.0

Time:1936

Location:Gaspé Peninsula, Quebec, Canada

Created by Dataset Agent

Overview

The Iris Dataset is the most iconic dataset in machine learning history—often called the "Hello World" of data science. Introduced by British statistician Ronald A. Fisher in his 1936 paper "The Use of Multiple Measurements in Taxonomic Problems," this multivariate dataset has become the universal standard for teaching classification algorithms and benchmarking new methods.

The dataset contains 150 samples of iris flowers, with 50 samples from each of 3 species: Iris setosa, Iris versicolor, and Iris virginica—a perfectly balanced dataset that's rare in real-world applications.

View Source

SQL

SELECT species, COUNT(*) AS count FROM iris.csv GROUP BY species ORDER BY species

Data

Species	Count
Iris-setosa	50
Iris-versicolor	50
Iris-virginica	50
3 rows

Understanding the Iris Flower

Before diving into the data, it's essential to understand what we're measuring. Many beginners encounter terms like "sepal" and "petal" without knowing what they mean botanically:

Sepals are the outer protective parts of the flower that enclose the bud before it blooms. In most flowers, sepals are green and leaf-like, but in iris flowers, they're colorful and often mistaken for petals. The three outer, drooping segments of an iris are actually sepals (called "falls" by gardeners).
Petals are the inner, typically colorful parts that attract pollinators. In iris flowers, these are the three upright segments (called "standards"). They're usually smaller than the sepals in iris species.
Length is measured from base to tip along the longest axis
Width is measured at the widest point perpendicular to the length

Unlike typical flowers where petals are larger than sepals, iris flowers have prominent sepals that are often more visually striking than the petals—making accurate measurement crucial for species identification.

Dataset Statistics by Species

The three species show dramatically different morphological characteristics, which is what makes this dataset excellent for classification tasks:

Summary Statistics: Sepal Length (cm) by Species

Species	Mean	Std Dev	Min	Max
Iris-setosa	5.01	0.35	4.3	5.8
Iris-versicolor	5.94	0.52	4.9	7.0
Iris-virginica	6.59	0.64	4.9	7.9
3 rows

View Source

SQL

SELECT species, ROUND(AVG(sepal_length), 2) AS mean, ROUND(STDDEV (sepal_length), 2) AS std_dev, ROUND(MIN(sepal_length), 1) AS min, ROUND(MAX(sepal_length), 1) AS max FROM iris.csv GROUP BY species ORDER BY mean

Data

Species	Mean	Std Dev	Min	Max
Iris-setosa	5.01	0.35	4.3	5.8
Iris-versicolor	5.94	0.52	4.9	7.0
Iris-virginica	6.59	0.64	4.9	7.9
3 rows

Summary Statistics: Petal Length (cm) by Species

Species	Mean	Std Dev	Min	Max
Iris-setosa	1.46	0.17	1.0	1.9
Iris-versicolor	4.26	0.47	3.0	5.1
Iris-virginica	5.55	0.55	4.5	6.9
3 rows

View Source

SQL

SELECT species, ROUND(AVG(petal_length), 2) AS mean, ROUND(STDDEV (petal_length), 2) AS std_dev, ROUND(MIN(petal_length), 1) AS min, ROUND(MAX(petal_length), 1) AS max FROM iris.csv GROUP BY species ORDER BY mean

Data

Species	Mean	Std Dev	Min	Max
Iris-setosa	1.46	0.17	1.0	1.9
Iris-versicolor	4.26	0.47	3.0	5.1
Iris-virginica	5.55	0.55	4.5	6.9
3 rows

The most striking difference: Iris setosa petal length averages just 1.46 cm, while Iris virginica averages 5.55 cm—nearly 4x larger. This dramatic gap makes setosa linearly separable from the other species.

View Source

SQL

SELECT species, ROUND(AVG(petal_length), 2) AS avg_petal_length FROM iris.csv GROUP BY species ORDER BY avg_petal_length

Data

Species	Avg Petal Length
Iris-setosa	1.46
Iris-versicolor	4.26
Iris-virginica	5.55
3 rows

Feature Correlations and Predictive Power

Not all features are equally useful for classification. Understanding which measurements matter most helps you build better models:

Feature Correlation Matrix

Feature Pair	Correlation	Predictive Value
Petal Length ↔ Petal Width	0.963	★★★★★ Excellent
Sepal Length ↔ Petal Length	0.872	★★★★☆ Very Good
Sepal Length ↔ Petal Width	0.818	★★★★☆ Very Good
Sepal Length ↔ Sepal Width	-0.109	★☆☆☆☆ Poor
Sepal Width ↔ Petal Width	-0.357	★★☆☆☆ Weak
5 rows

View Source

SQL

SELECT ROUND(CORR(petal_length, petal_width), 3) AS petal_corr, ROUND(CORR(sepal_length, petal_length), 3) AS sl_pl_corr, ROUND(CORR(sepal_length, sepal_width), 3) AS sepal_corr FROM iris.csv

Data

Feature Pair	Correlation	Predictive Value
Petal Length ↔ Petal Width	0.963	★★★★★ Excellent
Sepal Length ↔ Petal Length	0.872	★★★★☆ Very Good
Sepal Length ↔ Petal Width	0.818	★★★★☆ Very Good
Sepal Length ↔ Sepal Width	-0.109	★☆☆☆☆ Poor
Sepal Width ↔ Petal Width	-0.357	★★☆☆☆ Weak
5 rows

Petal measurements are far more predictive than sepal measurements for species classification. The 0.963 correlation between petal length and petal width means these features carry nearly identical information—you could use just one and lose minimal accuracy.

View Source

SQL

SELECT ROUND(CORR(petal_length, petal_width), 3) AS petal_correlation FROM iris.csv

Data

Petal Correlation
0.963
1 row

Pro tip: When building models, try using only petal_length and petal_width first. You'll often achieve 95%+ accuracy with just 2 features instead of 4, making your model simpler and more interpretable.

Linear Separability: The Key Insight

One of the most important characteristics of this dataset—and why it's used to teach classification—is its partial linear separability:

Iris setosa is linearly separable from versicolor and virginica. A simple rule like "petal_length < 2.5 cm" correctly identifies 100% of setosa samples.
Versicolor and virginica overlap in feature space. No straight line can perfectly separate them, requiring more sophisticated algorithms or accepting some misclassification.
This makes the dataset ideal for demonstrating why simple linear classifiers work sometimes but not always.

Using petal_length < 2.5 cm as a threshold, all 50 setosa samples are correctly isolated with zero false positives from the other 100 samples.

View Source

SQL

SELECT species, COUNT(*) AS count FROM iris.csv WHERE petal_length < 2.5 GROUP BY species

Data

Species	Count
Iris-setosa	50
1 row

Historical Context

The Iris dataset was introduced by Ronald A. Fisher in his 1936 paper "The Use of Multiple Measurements in Taxonomic Problems" published in the Annals of Eugenics. Fisher used this data to demonstrate linear discriminant analysis (LDA), a method he developed for classifying observations into predefined categories.

The actual measurements were collected by Edgar Anderson, an American botanist who gathered the data from iris flowers on the Gaspé Peninsula in Quebec, Canada. Anderson's meticulous measurements of 50 specimens from each of the three species provided Fisher with a perfectly balanced dataset for his statistical analysis. The collaboration between Anderson's fieldwork and Fisher's statistical methods exemplifies early interdisciplinary data science.

Known Data Quality Issues

The UCI Machine Learning Repository version contains two documented transcription errors that have propagated through many copies of this dataset.

The known errors are in samples 35 and 38 (0-indexed), both Iris setosa:

For most educational purposes, these minor errors don't significantly impact results. However, if you're publishing research or need exact reproducibility, consider using the corrected version from Fisher's original paper or noting which version you used.

Why This Dataset is Perfect for Beginners

The Iris dataset has specific properties that make it ideal for learning machine learning fundamentals:

Limitations for Real-World ML

While excellent for learning, the Iris dataset has significant limitations that don't reflect real-world machine learning challenges:

Too small: 150 samples is trivial—modern datasets have millions of records
Too clean: No missing values, no outliers, no noise—unrealistic for production data
Too balanced: Real classification problems often have 99:1 or worse class imbalance
Too easy: Most algorithms achieve 95%+ accuracy, making it hard to compare methods
Too simple for deep learning: Neural networks need thousands of samples to show their advantages over simpler methods

If your model achieves less than 90% accuracy on Iris, there's likely a bug in your code. If it achieves 100%, you may be overfitting or evaluating on training data. Target 95-98% accuracy as a sanity check.

Sample Data Preview

First 5 Records from Each Species

#	Sepal Length	Sepal Width	Petal Length	Petal Width	Species
1	5.1	3.5	1.4	0.2	Iris-setosa
2	4.9	3	1.4	0.2	Iris-setosa
3	7	3.2	4.7	1.4	Iris-versicolor
4	6.4	3.2	4.5	1.5	Iris-versicolor
5	6.3	3.3	6	2.5	Iris-virginica
6	5.8	2.7	5.1	1.9	Iris-virginica
6 rows

View Source

SQL

( SELECT * FROM iris.csv WHERE species = 'Iris-setosa' LIMIT 2 ) UNION ALL ( SELECT * FROM iris.csv WHERE species = 'Iris-versicolor' LIMIT 2 ) UNION ALL ( SELECT * FROM iris.csv WHERE species = 'Iris-virginica' LIMIT 2 )

Data

Sepal Length	Sepal Width	Petal Length	Petal Width	Species
5.1	3.5	1.4	0.2	Iris-setosa
4.9	3	1.4	0.2	Iris-setosa
7	3.2	4.7	1.4	Iris-versicolor
6.4	3.2	4.5	1.5	Iris-versicolor
6.3	3.3	6	2.5	Iris-virginica
5.8	2.7	5.1	1.9	Iris-virginica
6 rows

Expected Model Performance

Use these benchmarks to validate your implementations:

Table Overview

iris

Contains 150 rows and 5 columns. Column types: 4 numeric, 1 text.

150 rows5 columns

iris

150

rows

columns

Data Preview

Scroll to see more

sepal_length	sepal_width	petal_length	petal_width	species
5.1	3.5	1.4	0.2	Iris-setosa
4.9	3	1.4	0.2	Iris-setosa
4.7	3.2	1.3	0.2	Iris-setosa
4.6	3.1	1.5	0.2	Iris-setosa
5	3.6	1.4	0.2	Iris-setosa

Row 1

sepal_length5.1

sepal_width3.5

petal_length1.4

petal_width0.2

speciesIris-setosa

Row 2

sepal_length4.9

sepal_width3

petal_length1.4

petal_width0.2

speciesIris-setosa

Row 3

sepal_length4.7

sepal_width3.2

petal_length1.3

petal_width0.2

speciesIris-setosa

Showing 5 of 150 rows

Data Profile

150

rows

columns

100%

complete

36.6 KB

estimated size

Column Types

4 Numeric1 Text

Data Dictionary

iris

Column	Type	Description	Example
`sepal_length`	numeric	Numeric value (range: 4.6 - 5.1)	5.1, 4.9
`sepal_width`	numeric	Numeric value (range: 3 - 3.6)	3.5, 3
`petal_length`	numeric	Numeric value (range: 1.3 - 1.5)	1.4, 1.4
`petal_width`	numeric	Numeric value (range: 0.2 - 0.2)	0.2, 0.2
`species`	string	Species	"Iris-setosa", "Iris-setosa"