How anujtiwari1427/sentiment-analysis-project works
This repository analyzes survey responses about local brands, trains a sentiment classifier, and presents survey and prediction insights in a Streamlit dashboard. The main user flow is comment entry → hybrid sentiment inference → dashboard results. Offline, survey text is normalized and weakly labeled before TF-IDF training and model evaluation; a separate EDA branch produces survey dictionaries and visualizations. The dashboard and training stages are excerpted, but some artifact-loading and visualization wiring is documented only in the README, so those links cite it.
Main entry points: predict.py Hybrid inference · app.py Streamlit dashboard · analyze_data.py Survey EDA · train_model.py Training pipeline
Components 12
Survey analysis
- Survey responses
data/survey_responses.xlsx - Survey EDA
src/analyze_data.py - Survey visualizations
outputs
Sentiment modeling
- Text preparation
src/sentiment_utils.py - Training pipeline
src/train_model.py - TF-IDF features
src/train_model.py - Sentiment classifier
src/train_model.py
Prediction and dashboard
- Hybrid inference
src/predict.py - Streamlit dashboard
app.py
Persisted outputs
- Model artifacts
models - Training results
outputs
Other components
- Dashboard user
Connections 15
- Survey responses → Survey EDA: reads survey
src/analyze_data.py - Survey responses → Training pipeline: loads responses
src/train_model.py - Survey EDA → Text preparation: uses normalization
src/analyze_data.py - Training pipeline → Text preparation: uses preprocessing
src/train_model.py - Training pipeline → TF-IDF features: extracts features
README.md - TF-IDF features → Sentiment classifier: fits classifier
README.md - Sentiment classifier → Model artifacts: serializes model
src/train_model.py - Sentiment classifier → Training results: exports metrics
src/train_model.py - Survey EDA → Survey visualizations: generates charts
README.md - Dashboard user → Streamlit dashboard: enters comment
README.md - Streamlit dashboard → Hybrid inference: requests prediction
app.py - Hybrid inference → Text preparation: normalizes text
src/predict.py - Hybrid inference → Model artifacts: loads model assets
README.md - Hybrid inference → Streamlit dashboard: returns prediction
README.md - Survey visualizations → Streamlit dashboard: displays survey visuals
README.md