NLP · TF-IDF · Logistic Regression

Fake News Detection Using Machine Learning

Paste a news headline or article below. The trained classifier cleans the text, converts it into TF-IDF features and predicts whether the content reads as REAL or FAKE news — along with a confidence score.

0 words · 0 charactersMinimum 20 characters

Try a sample

The prediction, confidence score and the words that influenced the decision will appear here.

Live model snapshot

Algorithm
Logistic Regression
Features
591 TF-IDF terms
Training samples
80
Test accuracy
90.0%
F1-score (fake)
0.909

Pipeline

How It Works

Paste

Copy any news headline or article into the input area.

Preprocess

The text is cleaned, tokenized, stemmed and stop words are removed.

Vectorize

TF-IDF converts the cleaned text into a weighted numeric feature vector.

Classify

Logistic Regression predicts REAL or FAKE with a confidence score.

Why this project

Key Highlights

Instant

Sub-second prediction in the browser, no server needed.

Explainable

See the confidence and the words that influenced the decision.

Accurate

Trained on 44K+ articles and validated with precision, recall and F1-score.

Academic

Complete Python/Flask + scikit-learn reference included in /python.