Iris dataset

Presentation

This is one of the best known dataset to be found in the machine learning literature.

It was introduced by the statistician Ronald Fisher in his 1936 paper “The use of multiple measurements in taxonomic problems”, Annals of Eugenics. 7 (2): 179-188.

It contains 150 instances of iris plants:

• 50 Iris Setosa,

• 50 Iris Versicolour,

• 50 Iris Virginica.

Each instance is characterized by:

• its sepal length in cm,

• its sepal width in cm,

• its petal length in cm,

• its petal width in cm.

This dataset can be used for either clustering purposes or classification ones.

from __future__ import annotations

from numpy.random import default_rng

from gemseo import configure_logger
from gemseo import create_benchmark_dataset
from gemseo.post.dataset.andrews_curves import AndrewsCurves
from gemseo.post.dataset.parallel_coordinates import ParallelCoordinates
from gemseo.post.dataset.radviz import Radar
from gemseo.post.dataset.scatter_plot_matrix import ScatterMatrix

configure_logger()

rng = default_rng(1)

Load Iris dataset

We can easily load this dataset by means of the high-level function create_benchmark_dataset():

iris = create_benchmark_dataset("IrisDataset")

and get some information about it

iris

GROUP	parameters				labels
VARIABLE	sepal_length	sepal_width	petal_length	petal_width	specy
COMPONENT	0	0	0	0	0
0	5.1	3.5	1.4	0.2	0
1	4.9	3.0	1.4	0.2	0
2	4.7	3.2	1.3	0.2	0
3	4.6	3.1	1.5	0.2	0
4	5.0	3.6	1.4	0.2	0
...	...	...	...	...	...
145	6.7	3.0	5.2	2.3	2
146	6.3	2.5	5.0	1.9	2
147	6.5	3.0	5.2	2.0	2
148	6.2	3.4	5.4	2.3	2
149	5.9	3.0	5.1	1.8	2

150 rows × 5 columns

Manipulate the dataset

We randomly select 10 samples to display.

samples = rng.choice(len(iris), size=10, replace=False)

We can easily access the 10 samples previously selected, either globally

data = iris.get_view(indices=samples)

or only the parameters:

iris.get_view(group_names=iris.PARAMETER_GROUP, indices=samples)

GROUP	parameters
VARIABLE	sepal_length	sepal_width	petal_length	petal_width
COMPONENT	0	0	0	0
46	5.1	3.8	1.6	0.2
66	5.6	3.0	4.5	1.5
120	6.9	3.2	5.7	2.3
5	5.4	3.9	1.7	0.4
140	6.7	3.1	5.6	2.4
72	6.3	2.5	4.9	1.5
21	5.1	3.7	1.5	0.4
107	7.3	2.9	6.3	1.8
136	6.3	3.4	5.6	2.4
37	4.9	3.6	1.4	0.1

or only the labels:

iris.get_view(group_names="labels", indices=samples)

GROUP	labels
VARIABLE	specy
COMPONENT	0
46	0
66	1
120	2
5	0
140	2
72	1
21	0
107	2
136	2
37	0

Plot the dataset

Lastly, we can plot the dataset in various ways. We will note that the samples are colored according to their labels.

Plot scatter matrix

We can use the ScatterMatrix plot where each non-diagonal block represents the samples according to the x- and y- coordinates names while the diagonal ones approximate the probability distributions of the variables, using either an histogram or a kernel-density estimator.

ScatterMatrix(iris, classifier="specy", kde=True).execute(save=False, show=True)

[<Figure size 640x480 with 16 Axes>]

Plot parallel coordinates

We can use the ParallelCoordinates plot, a.k.a. cowebplot, where each samples is represented by a continuous straight line in pieces whose nodes are indexed by the variables names and measure the variables values.

ParallelCoordinates(iris, "specy").execute(save=False, show=True)

[<Figure size 640x480 with 1 Axes>]

Plot Andrews curves

We can use the AndrewsCurves plot which can be viewed as a smooth version of the parallel coordinates. Each sample is represented by a curve and if there is structure in data, it may be visible in the plot.

AndrewsCurves(iris, "specy").execute(save=False, show=True)

[<Figure size 640x480 with 1 Axes>]

Plot Radar

We can use the Radar plot

Radar(iris, "specy").execute(save=False, show=True)

[<Figure size 640x480 with 1 Axes>]