Balenciaga PR Scandal Analysis: Sentiment and Brand Perception

Overview

This repository analyzes over 2 million tweets from the Balenciaga PR scandal to explore public sentiment shifts and inform crisis management strategies.

Key Features

• Data Collection: Tweets collected using the Twitter API and preprocessed by cleaning, translation, and filtering.
• Sentiment Analysis: Leveraged VADER and Text2Emotion to classify emotions like anger, sadness, and happiness.
• Quantitative Analysis: Applied t-tests and Regression Discontinuity Design (RDD) to evaluate PR impact and explore cultural differences.

---

Repository Structure

• scarping_tweets.ipynb: Code for collecting tweets using the Twitter API.
• translation.py: Script for translating non-English tweets.
• merging_translated_parts.py: Script for merging translated tweet files into a single dataset.
• process_sentiment.py: Script for cleaning tweets, analyzing sentiment, and extracting emotions.
• RDD_data_analysis.ipynb: Notebook for analyzing sentiment shifts and performing Regression Discontinuity Design (RDD).

---

Key Findings

• Apologies tend to increase negative emotions like sadness and anger temporarily.
• Discount campaigns improve positive sentiment and surprise in the long term.

---

Contact

For questions and data, please reach out via deryadurmush@gmail.com.