Difference between revisions of "Python-for-Machine-Learning/C2/Logistic-Regression-MultiClass-Classification/English"

Welcome

Learning Objectives

• Multiclass Classification for Logistic Regression

System Requirements

• Ubuntu Linux OS version 24.04
• Jupyter Notebook IDE

Prerequisite

• The learner must have basic knowledge of Python.
• For pre-requisite Python tutorials, please visit this website.

Code files

• The files used in this tutorial are provided in the Code files link.
• Please download and extract the files.
• Make a copy and then use them while practicing.

Iris flower classification

• Use the iris dataset to classify the iris flower.
• To know more about the iris dataset please watch earlier tutorials.

Type conda activate ml Press Enter Highlight: (ml)

Activate the machine learning environment as shown.

Type jupyter notebook

Press Enter

Please navigate to the respective folder of your code file location.

Then type, jupyter space notebook and press Enter.

Double Click on LR_Multiclass.ipynb file

Click on the LR underscore Multiclass dot ipynb file to open it.

Note that each cell will have the output displayed in this file.

Let us see the implementation of multiclass logistic regression.

Highlight: iris = load_iris()

iris.data[:5]

The dataset is stored in the variable iris.

Then we display the first five rows using the head method.

X = iris.data

df['target'] = Y

Now, we visualize this correlation using a heatmap.

The heatmap shows how features relate to one another.

mlr.fit(X_train, Y_train)

Set multi underscore class equals multinomial and solver equals lbfgs.

We also set max underscore iter equals 1000 to ensure convergence.

Now we train the model using the fit method on the training data.

Ignore the warning in the output cell, if any.

Y_train_pred = mlr.predict(X_train)

Train Log Loss: 0.1308

A Loss of 0.1308 shows the model is making accurate predictions.

plt.figure(figsize=(8, 6))

for i in range(Y_train_pred_proba.shape[1]): # Iterate over each class

fpr, tpr, _ = roc_curve(Y_train == i, Y_train_pred_proba[:, i]) # One-vs-rest for each class

roc_auc = auc(fpr, tpr)

The ROC curve shows TPR vs FPR at various threshold values.

TPR stands for True Positive Rate that is recall.

It is the fraction of actual positives correctly identified.

FPR stands for False Positive Rate.

It is the fraction of actual negatives wrongly classified as positives.

The curves stay close to the top-left corner.

All three classes achieve an AUC of 1.00.

This indicates the model effectively distinguishes the classes.

predicted_class = mlr.predict(test_data)

We compare the predicted with the actual test class.

The actual value is also 0, hence prediction is correct.

print(f"Test Accuracy: {test_accuracy:.3f}")

# Predict probabilities for test set

Y_test_pred_proba = mlr.predict_proba(X_test)

Test ROC-AUC Score (OvR): 0.9968

Test Log Loss: 0.1616

Cross-entropy loss of 0.1616 shows the predictions are accurate.

conf_matrix = confusion_matrix(Y_test, Y_pred)

It shows how well the model classifies each class.

The diagonal values represent correct predictions.

One sample from Class 1 was incorrectly predicted as Class 2.

The absence of other misclassified values indicates that the model performs well.

A strong diagonal pattern suggests high classification accuracy.

This brings us to the end of the tutorial. Let us summarize.

Summary

Assignment

• Generate the classification report of the model using sklearn method as shown.
• Use classification_report from sklearn dot metrics to display results.
• This shows precision, recall, f1-score, and support for each class.

FOSSEE Forum

Machine Learning, visit the FOSSEE forum and post your question.

Thank You

Thanks for joining.