Global Health Trends Extraction (COVID-19 Twitter Analysis)

This project performs end-to-end health trend analysis using classical NLP methods (no transformers). It processes COVID-19-related tweets to extract sentiment patterns, keyword trends, locations, entities, and time-series insights.

All code and visuals come from the notebook global_health_trends_extraction.ipynb (Colab).

---

Features

1. Data Loading

• Loads the Corona NLP dataset (CSV) via pandas.
• Includes columns: Username, Location, Date, Tweet Text, Sentiment.

2. Data Cleaning

A custom clean_text() function removes:

• URLs
• Mentions (@user)
• Hashtags
• HTML codes
• Extra spaces
  (pages 1–2) :contentReference[oaicite:1]{index=1}

3. Sentiment Analysis (VADER)

• Uses NLTK VADER to compute compound sentiment scores.
• Converts scores → Positive, Negative, Neutral.
• Generates sentiment-trend line graphs.
  (pages 2–3 & 7–8) :contentReference[oaicite:2]{index=2}

4. Keyword Extraction (YAKE)

• Extracts top keywords for each tweet.
• Aggregates and plots Top 20 keywords.
  (page 4 chart) :contentReference[oaicite:3]{index=3}

5. Named Entity Recognition (spaCy)

• Uses en_core_web_sm.
• Extracts all entities per tweet.
• Filters locations only (GPE, LOC).
• Plots top user locations & top mentioned locations.
  (pages 5–7 visuals) :contentReference[oaicite:4]{index=4}

6. Time-Based Trend Analysis

• Tweet volume per day
• Sentiment trend by date
  (page 7–8) :contentReference[oaicite:5]{index=5}

---

Technologies Used

Purpose	Technology	Reference
Data Loading	pandas	page 1
NLP Dataset	HuggingFace CSV	page 1
Cleaning	re (regex)	pages 1–2
Sentiment Analysis	nltk (VADER)	pages 2–3
Keyword Extraction	yake	pages 3–4
NER	spaCy	page 5
Visualization	matplotlib	all graphs
Date Parsing	pandas datetime	page 2

(All validated from notebook.) :contentReference[oaicite:6]{index=6}

---