bolubme/Geo-semantic-analysis

This project evaluates the service quality of hospitals in Nigeria by analyzing citizens’ opinions expressed on social media. # Geo-semantic-analysis Service Quality of Hospitals in Nigeria: Geo-Semantic Analysis of Citizens' Social Media Posts Author: Morolari Boluwatife Organization: Data Science Nigeria, Lagos, Nigeria Email: morolaribolu@gmail.com 📖 Overview This project evaluates the service quality of hospitals in Nigeria by analyzing citizens’ opinions expressed on social media. Using Twitter data, combined with geo-semantic analysis, natural language processing (NLP), and sentiment analysis, this study provides insights into: Public perception of hospital services Geographic variations in hospital satisfaction Key factors contributing to service quality issues The results can inform policymakers, healthcare providers, and stakeholders to improve service delivery. 🎯 Objectives Assess citizens' sentiment towards hospital services in Nigeria. Identify critical factors affecting service quality (e.g., infrastructure, waiting times, staff professionalism). Map geographical variations in hospital service perception. Provide actionable insights for healthcare policy and hospital management. 🛠 Methods 1. Data Collection Platform: Twitter Tool: Twint (Python-based scraper, no API required) Keywords: Names of hospitals across Nigeria Data Collected: Tweet ID, timestamp, content, author handle 2. Data Preprocessing Removed irrelevant tweets (e.g., <3 words, unrelated mentions) Cleaned text (hyperlinks, hashtags, mentions, emojis, numbers, punctuation) Tokenization & Lemmatization: Using NLTK Removed stop words for meaningful analysis Final Dataset: 250 curated tweets 3. Data Analysis Topic Modeling Used Short Text Topic Modeling (STTM) with GSDMM (Gibbs Sampling Dirichlet Mixture Model) Identified key topics: Infrastructure & Facilities Staff Behavior & Professionalism Waiting Times & Appointment Management Accessibility & Availability Overall Patient Experience Sentiment Analysis Library: TextBlob Sentiment polarity: -1 (negative) to +1 (positive) Geo-Sema …