Iris Classification with Machine Learning: Data Exploration and Classic Algorithms

In recent years, artificial intelligence (AI) technologies have surged, with OpenAI releasing products such as ChatGPT and Sora. This article guides beginners through AI developments using the classic Iris classification dataset.

Dataset Introduction

The Iris dataset contains 150 samples of three species—Setosa, Versicolor, and Virginica—each described by four features: sepal length, sepal width, petal length, and petal width.

Data can be loaded with pandas:

import pandas as pd

iris = pd.read_csv('./iris.csv', names=['sepal_length','sepal_width','petal_length','petal_width','class'])

print(iris.head(10))

Exploratory Data Analysis

Descriptive statistics, histograms, KDE plots, and correlation heatmaps are used to understand feature distributions and relationships.

iris.describe()

iris.plot(kind='hist', subplots=True, layout=(2,2), figsize=(10,10))

iris.plot(kind='kde')

sns.heatmap(iris.iloc[:,:4].corr(), annot=True, cmap='YlGnBu')

Classification Algorithms

Four classic classifiers are demonstrated: Decision Tree (CART), Logistic Regression, Support Vector Machine, and K‑Nearest Neighbors.

Decision Tree

Model training:

from sklearn import preprocessing, model_selection, tree

label_encoder = preprocessing.LabelEncoder()

target = label_encoder.fit_transform(iris['class'])

X_train, X_test, y_train, y_test = model_selection.train_test_split(iris.iloc[:,:4].values, target, test_size=0.2, random_state=42)

clf = tree.DecisionTreeClassifier(max_depth=4)

clf.fit(X_train, y_train)

print(clf.feature_importances_)

Visualization with Graphviz:

import pydotplus

dot_data = tree.export_graphviz(clf, out_file=None, feature_names=['sepal_length','sepal_width','petal_length','petal_width'], class_names=iris['class'].unique(), filled=True, rounded=True)

graph = pydotplus.graph_from_dot_data(dot_data)

graph.write_png('decision_tree.png')

Logistic Regression

Training and evaluation:

from sklearn.linear_model import LogisticRegression

model = LogisticRegression()

model.fit(X_train, y_train)

y_pred = model.predict(X_test)

accuracy = metrics.accuracy_score(y_test, y_pred)

Support Vector Machine

Using an RBF kernel and visualizing decision regions:

from sklearn import svm

model = svm.SVC(kernel='rbf', gamma=10, C=10.0, random_state=0)

model.fit(X_train_std, y_train)

# plot_decision_regions function omitted for brevity

K‑Nearest Neighbors

Training and plotting decision boundaries:

from sklearn.neighbors import KNeighborsClassifier

knn = KNeighborsClassifier(n_neighbors=2, p=2, metric='minkowski')

knn.fit(X_train_std, y_train)

Evaluation metrics such as accuracy, precision, recall, and F1 score are reported, often achieving 100 % on the test split due to the simplicity of the dataset.

Conclusion

The article demonstrates a complete workflow—from data loading and exploratory analysis to model training and visualization—for the Iris classification problem, providing a practical entry point for beginners in AI and machine learning.