How rsd018/prediction-des-recommandations--dachats--de-vetements--pour--femmes works
This project turns customer clothing reviews into a binary recommendation classifier and exposes the saved model through a Streamlit dashboard. Its documented workflow runs from dataset cleaning through feature preparation and model selection to live prediction. The README describes the notebook stages and dashboard capabilities; source excerpts confirm model-resource loading and dashboard navigation, but do not expose the prediction-call implementation, so that wiring is based on README documentation.
Main entry points: app.py Streamlit dashboard · 02_preprocessing_encoding.ipynb Feature preparation · 03_classification_models.ipynb Model comparison · preprocessor.pkl Saved preprocessor
Components 18
Data preparation
- Raw reviews
data/raw/reviews.csv - EDA and cleaning
notebooks/01_eda_and_cleaning.ipynb - Clean reviews
data/processed/reviews_clean.csv - Feature preparation
notebooks/02_preprocessing_encoding.ipynb
Model training
- Saved preprocessor
models/preprocessor.pkl - Review TF-IDF
models/review_tfidf.pkl - Title TF-IDF
models/title_tfidf.pkl - Model comparison
notebooks/03_classification_models.ipynb - Linear SVC model
models/best_model.pkl
Interactive application
- Streamlit dashboard
src/app.py - Review and product details
src/app.py - Recommendation result
src/app.py - Model performance
src/app.py - Dataset statistics
src/app.py - Key findings
src/app.py - Project information
src/app.py
Other components
- Project analyst
- Dashboard user
Connections 21
- Project analyst → Raw reviews: supplies dataset
- Raw reviews → EDA and cleaning: reads and cleans
- EDA and cleaning → Clean reviews: saves cleaned data
- Clean reviews → Feature preparation: feeds preparation
- Feature preparation → Saved preprocessor: saves transformer
- Feature preparation → Review TF-IDF: saves vectorizer
- Feature preparation → Title TF-IDF: saves vectorizer
- Saved preprocessor → Model comparison: transforms features
- Review TF-IDF → Model comparison: vectorizes reviews
- Title TF-IDF → Model comparison: vectorizes titles
- Model comparison → Linear SVC model: saves selected model
- Dashboard user → Streamlit dashboard: uses dashboard
- Streamlit dashboard → Saved preprocessor: loads and applies
- Streamlit dashboard → Linear SVC model: loads and predicts
- Dashboard user → Review and product details: enters review
- Review and product details → Streamlit dashboard: submits input
- Streamlit dashboard → Recommendation result: shows prediction
- Streamlit dashboard → Model performance: displays comparison
- Streamlit dashboard → Dataset statistics: displays distributions
- Streamlit dashboard → Key findings: displays findings
- Streamlit dashboard → Project information: displays project details