BCF1 Group 6 - Fake News Detection Repo

Information

• Repository for SC1015 Intro to DSAI AY21/22 Sem 2 project
• Application of the Machine Learning Pipeline in the order of

1. Data Extraction & Preparation
2. Exploratory Data Analysis
3. Machine Learning Models
4. Evaluation and Insights

Team Members

• Kane Tan (Data Extraction & Preparation)
• Javier Tan (Data Visualisation & Analysis)
• Lee Seung Soo (Machine Learning)

Problem Statement

Predicting fake news by the news title using Natural Language Processing

Models Used

Ensemble model consisting of 5 sub-models

1. Logistic Regression
2. Naives-Bayes Classifer
3. Binary Tree Classifier
4. Passive Aggressive Classifier
5. Support Vector Machine

Outcome

• NLP is a good way of classifying news by the title
• High prediction accuracy of 83.5% of the ensemble model
• Good warning system for users' exposure to fake news online
• Cross-checking about the content of the news is still required

References

https://www.kaggle.com/datasets/clmentbisaillon/fake-and-real-news-dataset https://www.datacamp.com/community/tutorials/understanding-logistic-regression-python https://www.datacamp.com/community/tutorials/svm-classification-scikit-learn-python https://www.datacamp.com/community/tutorials/ensemble-learning-python https://arxiv.org/pdf/2102.04458#:~:text=Machine%20learning%20classifiers%20are%20using,can%20automatically%20detect%20fake%20news.