frankraDIUM/Sentiment-Analysis-X-Twitter-data

Sentiment analysis of flood risk perception in the Greater Accra Region of Ghana using X(Twitter) data --- # Sentiment Analysis of Flood Risk Perception in Greater Accra using X/Twitter Data ## Project Overview This project performs **Sentiment Analysis** on X/Twitter data to gauge public perception and emotional tone regarding flood risk in Ghana's Greater Accra Region. By scraping and analyzing tweets containing flood-related keywords, this study categorizes public sentiment into **Positive, Negative, and Neutral** classes. The goal is to provide insights into community concerns, government response perceptions, and overall social vulnerability, complementing traditional geospatial flood risk models with human-centric data. ## Study Area Context The analysis focuses on public discourse surrounding flooding in the **Greater Accra Region**, Ghana's capital and most densely populated area. This region is highly vulnerable to flooding due to its: * Low-lying coastal plains * Rapid urbanization * Inadequate drainage infrastructure * Frequent heavy rainfall events Understanding public sentiment here is crucial for effective disaster communication, policy-making, and community engagement strategies. ## Data Collection * **Source:** X (formerly Twitter) * **Method:** Data was scraped using third-party tools (Twitter Scraper & Twitter Scraper V2) on Apify.com. * **Keywords:** Tweets were collected using flood-related buzzwords and specific location names (e.g., `"Weija flood"`, `"Adabraka flood"`, `"Tse Addo flood"`, `"Kaneshie flood"`, `"Odaw drain"`). * **Dataset:** 1,232 publicly accessible user tweets were collected and downloaded in CSV format. * **Data Fields:** The dataset includes: * `tweet` (text content) * `author username`, `location`, `description` * `date` of tweet * Engagement metrics (`likes count`, `retweets`) * Author account details (`followers`, `following`, `account creation date`) ## Methodology & Processing ### 1. Data Preprocessing Raw tweet data contains noise that must be cleaned for accurate analysis. The fol …