This repository contains a sentiment analysis report generated from predicted sentiments made by a model trained to analyze reviews of Cuiabá IT Companies' in Glassdoor. The model harnesses the power of BERT-based architecture, specifically BERTimbau, to predict the sentiment of reviews.
The live preview can be found at: Análise de sentimento em avaliações no Glassdoor: Um estudo sobre empresas de Tecnologia da Informação em Cuiabá.
Serves as the primary access point for the application Loads the necessary data for analysis and implements the following sections:
- Introduction: Provides an overview of the report.
- Positive and Negative Reviews Ranking: Analyzes and ranks reviews based on sentiment.
- Company Analysis: Evaluates company reputation.
- Sentiment Reviews Over Time: Tracks sentiment trends across different time periods.
- Rating Star Analysis: Examines distribution and patterns of rating stars.
- Employee Role Analysis: Investigates reviews based on employee roles.
- Word Cloud Analysis: Visual representation of frequently used words in reviews.
- Most Common Words Analysis: Lists the most common words in reviews.
- N-Gram Analysis: Analyzes most frequent sequences of words.
Shows the number of reviews and the associated sentiment for each company, ordered by the difference between positive and negative reviews.
Presents a sentiment analysis of reviews along the time by company.
Examines how reviews correlate with star ratings for a chosen company.
Shows reviews categorized by different employee groups for a specific company.
Shows Word Cloud by company.
Shows Top 10 frequent words by company.
Shows N-Grams by company.
Shows Data Preparation, Model Architecture, Model Training and Model Evaluation.
-
Clone the repository:
git clone https://github.com/yourusername/glassdoor-reviews-report.git cd glassdoor-reviews-report -
Install the required dependencies:
pip install -r requirements.txt
-
Run the Streamlit application:
streamlit run 🏠Home.py