How naak-ktr/sentiment-analysis-detecting-food-insecurity-risk-twitter works
Project type and purpose
• This is a data-science pipeline for sentiment analysis of Twitter data to detect food insecurity risk.
• It is neither a classic full-stack web app nor a microservice; instead it is an ETL + analytics + reporting system.
File structure
• Top-level folders represent pipeline phases:
• Data Collection
• Data Preprocessing & Cleaning
• Data Visualization
• Model Development & Evaluation
• dataset (raw and cleaned CSVs)
• Notebooks (.ipynb) implement each phase.
• A Power BI file (.pbix) and dashboard link serve as the front-end reporting component.
• External tools invoked: Python (Snscrape), RapidMiner.
README for insights
• Stepwise method:
1) Data Collection & Preprocessing
2) Lexicon-Based Sentiment Scoring (TextBlob, VADER, Harmonized)
3) Risk classification via ML (SVM, NB, LR)
4) Model evaluation (70:30, 80:20 splits)
5) Dashboard visualization
Details
• Break the “dataset” component into Raw vs. Cleaned sub-blocks.
• Three lexicon scoring branches merging into the harmonized pipeline.
• Finally, route the winning model output to the dashboard.