Top 5 Data Science Capstone Project Ideas That Will Impress Employers and Sharpen Your Skills

📗 Chapter 3: Fake News Detection Using Natural Language Processing

Build a Machine Learning Model to Separate Facts from Fabrication

🧠 Introduction

The internet has transformed how we consume news, but it's also given rise to a dangerous byproduct — fake news. From political misinformation to health-related hoaxes, fake news has the power to influence public opinion and cause real-world harm.

This capstone project teaches you how to use Natural Language Processing (NLP) to build a model that detects fake news articles using text classification techniques.

In this chapter, you'll learn:

• The basics of text preprocessing
• NLP feature extraction with TF-IDF
• Building classifiers like Naive Bayes and SVM
• Model evaluation with accuracy and confusion matrix
• Extensions using word embeddings and explainability

By the end, you’ll have a fake news classifier ready to deploy or demonstrate in your data science portfolio.

📦 Step 1: Understanding the Problem

• Objective: Classify news articles as FAKE or REAL
• Dataset: Text-based articles labeled with their authenticity
• Impact: Can be used by media houses, social media platforms, or browser plugins

📊 Step 2: Load the Dataset

We’ll use the Fake and Real News Dataset from Kaggle.

📥 Dataset: Fake and Real News Dataset

python

import pandas as pd

df_fake = pd.read_csv("Fake.csv")

df_real = pd.read_csv("True.csv")

# Add labels

df_fake['label'] = 0

df_real['label'] = 1

df = pd.concat([df_fake, df_real]).sample(frac=1).reset_index(drop=True)

df.head()

🧹 Step 3: Text Cleaning & Preprocessing

python

import re

import string

def clean_text(text):

    text = text.lower()

    text = re.sub(r'https?://\S+', '', text)           # remove links

    text = re.sub(r'\[.*?\]', '', text)                # remove brackets

    text = re.sub(f"[{re.escape(string.punctuation)}]", '', text)

    text = re.sub(r'\w*\d\w*', '', text)               # remove words with numbers

    return text

df['text'] = df['title'] + " " + df['text']  # Combine title and body

df['text'] = df['text'].apply(clean_text)

📚 Step 4: Feature Extraction with TF-IDF

python

from sklearn.feature_extraction.text import TfidfVectorizer

vectorizer = TfidfVectorizer(stop_words='english', max_df=0.7)

X = vectorizer.fit_transform(df['text'])

y = df['label']

🔁 Step 5: Train-Test Split

python

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

🤖 Step 6: Build Classification Models

1. Naive Bayes

python

from sklearn.naive_bayes import MultinomialNB

from sklearn.metrics import accuracy_score, confusion_matrix, classification_report

nb_model = MultinomialNB()

nb_model.fit(X_train, y_train)

y_pred_nb = nb_model.predict(X_test)

print("Accuracy:", accuracy_score(y_test, y_pred_nb))

2. Support Vector Machine (SVM)

python

from sklearn.svm import LinearSVC

svm_model = LinearSVC()

svm_model.fit(X_train, y_train)

y_pred_svm = svm_model.predict(X_test)

print("Accuracy:", accuracy_score(y_test, y_pred_svm))

📈 Step 7: Model Evaluation

python

import seaborn as sns

import matplotlib.pyplot as plt

cm = confusion_matrix(y_test, y_pred_svm)

sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=['Fake', 'Real'], yticklabels=['Fake', 'Real'])

plt.xlabel('Predicted')

plt.ylabel('True')

plt.title('Confusion Matrix')

plt.show()

📋 Classification Report

python

print(classification_report(y_test, y_pred_svm, target_names=['Fake', 'Real']))

🔍 Step 8: Feature Importance (Optional for SVM)

python

import numpy as np

feature_names = vectorizer.get_feature_names_out()

coefs = svm_model.coef_.flatten()

top_positive_coefficients = np.argsort(coefs)[-20:]

top_negative_coefficients = np.argsort(coefs)[:20]

top_words = [feature_names[i] for i in top_positive_coefficients]

top_fake_words = [feature_names[i] for i in top_negative_coefficients]

print("Top Real-indicative words:", top_words)

print("Top Fake-indicative words:", top_fake_words)

🌐 Step 9: Model Deployment Ideas

• Streamlit App: Enter headline/text and get prediction
• Flask API: For use in web plugins or CMS
• Real-time dashboard: Detect fake news across headlines

📋 Summary Table