The Project
Overview
Culinary data analysis with complete workflow: exploration → development → visualization → deployment.
Development Process
1. Data Exploration (00_eda/)
Jupyter notebooks for exploratory analysis
9 notebooks organized by theme (trends, seasonality, weekend, ratings)
Pattern and insight identification
2. Application Development (10_preprod/)
Transform notebook analyses → reusable Python modules
Streamlit application for interactive result presentation
Modular architecture (utils, visualization, data, exceptions)
3. Deployment (20_prod/)
Automated CI/CD pipeline
Separate preprod/prod environments
Continuous testing and validation
EDA → Application Mapping
Each Streamlit analysis is based on one or more EDA notebooks:
Trend analysis ← 01_long_term/recipe_analysis_trendline.ipynb
Seasonality analysis ← 02_seasonality/recipe_analysis_seasonality.ipynb
Weekend analysis ← 03_week_end_effect/recipe_analysis_weekend.ipynb
Ratings analysis ← 01_long_term/rating_analysis.ipynb
The notebooks contain complete exploration (descriptive statistics, visualizations, hypothesis testing). The Streamlit application presents results interactively.
Technical Stack
Python 3.13.7
Streamlit 1.50.0 (web interface)
DuckDB 1.4.0 (OLAP database)
Polars 1.19.0 (data processing)
Plotly 5.24.1 (visualizations)
Environments
PREPROD: https://mangetamain.lafrance.io/ (port 8500)
PRODUCTION: https://backtothefuturekitchen.lafrance.io/ (port 8501)
Applied Standards
Complete type hinting
Pytest unit tests (93% coverage)
Google-style docstring documentation
Automatic PEP8 validation
Custom exception handling
CI/CD pipeline with GitHub Actions