Spam Detection – Interactive UI

Prior Probability Control

Drag either slider to modify class priors and observe how it affects posterior probabilities and accuracy of a Naive Bayes classifier.

Bayes' Theorem

\[P(C \mid X) = \frac{P(X \mid C)\,P(C)}{P(X)}\]

Posterior ∝ Likelihood × Prior — we adjust P(C) with the sliders below

Ham Prior Probability

0.82

Spam Prior Probability

0.18

Original Class Distribution

Posterior Probabilities \(P(C\mid X)\)

Summary

P(Ham) —

P(Spam) —

Accuracy —

Prediction —

Select a message (or choose Custom), click Analyse, and adjust Top N words to see feature contributions.

Naive Bayes Independence Assumption

\[P(X \mid C) = \prod_{i=1}^{n} P(x_i \mid C)\]

Each word contributes independently — multiply all individual likelihoods

Message

Top N words

Custom message

Individual Feature Likelihoods

Posterior Probabilities

P(Ham | X) —

P(Spam | X) —

Predicted Class —

Train and compare Logistic Regression, KNN, and Naive Bayes on the same TF‑IDF features.

Not trained

Click the button to train models and view metrics.

Test Sample

Classifiers Comparison on the selected sample Ham Spam

Logistic Regression

K‑Nearest Neighbors

Naive Bayes