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.
Try a sample
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.